-

Forse u gszentrum Ka sTechnik nd Umwelt

FZKA 5596

Compatibility of31 Metals, Alloys andCoatings with StaticPb...17li Eutectic Mixture

H. Feuerstein, H. Gräbner, J. Oschinski,J. Beyer, S. Horn, l. Hörner, K. SantoHauptabteilung IngenieurtechnikProjekt Kernfusion

September 1995

Forschungszentrum KarlsruheTechnik und Umwelt

Wissenschaftl iche Berichte

FZKA 5596

Compatibility of 31 Metals, Alloys

and Coatings with Static

Pb...17li Eutectic Mixture

H. Feuerstein, H. Gräbner, J. Oschins"ki, J. Beyer, S. Horn,L. Hörner and K. Santo

Hauptabteilung Ingenieurtechnik

Projekt Kernfusion

FORSCHUNGSZENTRUM KARlSRUHE GmbH, KARlSRUHE

1995

Compatibility of 31 metals. alloys and coatings with StaticPb-17li Eutectic Mixture

Abstract

The eompatibility of 31 metals, alloys and eoatings with statie euteetie mixturePb-17li was investigated in more than 3ee tests. Most of the results have not beenpublished before.

Wetting has no influenee on dissolution rates. This is diseussed in detail. Metalsean be divided into three groups. Most stable are the refraetories Nb, Ta, Mo, Reand W. Ferritie steels, Be, Fe, and V belong to the next group. However, Be isdestroyed along grain boundaries. Not stable at all are Al, Ti, Zr, Y, U and theiralloys.

Temperature funetions for solubilities in Pb-Ini were obtained for 8 elements,single -one temperature- values for 3 others. The results are in good agreementwith a theoretieal work of Guminski. Remarkable high are solubilities of Al, Zr, Yand U while those of the refraetories are low. Also, the solubility of Pb in solidTi was determined, adding new data points to the phase diagram.

Beeause of the effeet of mass transfer between dissimilar metals, diffusion coeffi­eients in Pb-17li eould be ealeulated from dissolution rates and solubilities. Mostreliable are the temperature funetions for Be, Al, Fe and V. Those for Ti, Zr andU are influeneed by the formation of eompounds. All values are in an expeetedrange, but not all effeets ean be explained.

Different kinds of reaction zones were found on surfaees. New is a very thin 'ehe­mieal reaetion zone', identified for several metals during sample eleaning. It isprobably formed as a first step before grain boundary attaek of the euteetie.

The following new intermetallic compounds were identified: TizPb and Ti 3Pbz, UPb~,

YPb~ and Ir~Pb. The eompound ThPb z was investigated in detail. lead and titaniumean be replaeed by other metals. With Y and U, there was even a reaetion with leadin the gas phase above the euteetie. Other metals were embrittled in this area.

Generally, alloys are not more stable than their base metals. leaehing of elementsfrom alloys and other effeets were investigated. Espeeially with alloys, many openquestions remain and more work has to be done to understand the ehemistry of alloysin the euteetie.

last but not least Mo coatings on getter metals were found not to be proteetive forthe use in a blanket.

Content

Abstract

Zusammenfassung

Preface

Part-I General considerations and results

1. Introduction

2. Experiments

3. Basic considerations

4. Results and discussion

5. Summary

Part-lI Results

Part-III Description of experiments

1. Used Materials and preparation

2. Facility and experiments

3. Chemistry and analysis

4. Evaluation of results

Appendix

References

list of companies

Acknowledgement

1

3

3

4

9

27

29

133

135

137

141

147

149

151

159

161

1

Part-I : General considerations and results

1- Introduction 3

2. Experiments 3

3. Basic considerations 4

3.1 Wetting 43.2 Dissolution 53.3 Mass transfer, diffusion coefficients 63.4 Surface reaction Iones 63.5 Intermetallic compounds 63.6 Mutual solubilities 13.1 Gas phase reactions 13.8 Complex mechanisms 13.9 Some remarks on molybdenum crucibles 1

4. Results and discussion 9

4.1 Wetting by Pb-11Li 94.2 Dissolution and mass transfer 94.3 Solubilities of elements in Pb-11Li 134.4 Diffusion coefficients in Pb-11Li 184.5 Reaction Iones and compounds 21

4.5.1 Chemical reaction Iones4.5.2 Metallographical reaction Iones4.5.3 Chemical compounds on surfaces4.5.4 Chemical compounds in solution

4.6 Behaviour of alloys 234.1 Mo-coatings 244.8 Solubilities of Li and Pb in solid metals 244.9 Gas phase reactions 26

5. Summary 21

1. Introduction

The eutectic mixture Pb-17Ul is proposed as one possible blanket material forfusion reactors (1). Physical and chemical properties of this mixture are describedin references (2) to (4). The eutectic contains 15.7 at.% lithium, its meltingpoint is 234 ee (5,6,7).

In a fusion reactor the mixture will be in contact with different materials. Theknowledge until 1998 about compatibilities is compiled by T.Sample (6). A number ofcorrosion experiments with steels have been done during the last years.

Because of the low chemical activity of Li in the eutectic it is assumed that mostproperties of the mixture will be similar to those of lead (3). This is only partlytrue. Solubilities in the eutectic of Mn, Ni, Th and Sm are up to a factor fourhigher (8,9). On the other hand many effects are comparable for different liquidmetals. This was shown e.g. by Guminski (Je) or Tortorelli (11). In this paper,results are often compared with those of pure lead. For some metals also resultsfrom liquid lithium experiments are included.

It was not possible to evaluate the literature for each investigated metal comple­tely, but attempts were made to find most references. The start was always phasediagram considerations on the basis of Massalski 's compilation (12). Furthermoresome phase diagrams were picked from 'Bulletin Alloy Phase Diagrams'(13) and fromthe ASM monograph serie (14). The second starting point was the mentioned thesis ofT.Sample (6).

2. Experiments

Details of the experiments are given in Part-III of this report.

The used materials were cleaned and/or vacuum annealed. The used eutectic mixturecontained between 15.5 and 17.8 at.% Li (8.61 to 0.72 wt.%). Because of segregationeffects (7) prior to the filling of crucibles, the concentration was not constant2•

In an Argon filled glovebox, 68 to 80 grams of the eutectic were filled into Mocrucibles. 3 15 cm2 Mo were wetted by the eutectic.

Sheets of sample materials were inserted. The sample surface in contact with theeutectic was 3 to 8 cm2 • Any contact between the sample metal and the crucible wallwas avoided. The assembly was then transferred - still inside the box -

1 The older value for the eutectic composition was 17 at.% Li. Most of the experiments weredone with thi s concentrat ion. However, the true Li concentrat ion of the eutect icis 15.7at.% (5,7). Nevertheless, the writing Pb-17Li will be used in this report.

2 No influence of the Li concentration on results was seen. The Li concentration will not bementioned any more.

3 In a few experiments crucibles of Fe, Wor alumina were used.

5

te of its low chemical activity Li is 11 a reducing for many oxides.Dissolution is accelerated after the removal of these des. Other authors did notfind such an incubation period with steels (11.18).

The influence of wetting on corrosion processes is not completely understood. Themodel with a gap between the solid and the liquid is too simple. Robertson (19.20)found that the vo1ume-diffusi on coeHi ci ents for Fe and Cr in 1ead were the samefor wetted and non-wetted surfaces. He observed 'grooving' at grain boundaries alsowhen metals were heated in vacuum or an inert gas. Groovi ng in a 1i qui d meta1surrounding is caused by volume diffusion in the liquid; grooving in vacuum or aninert gas by diffusion in the solid metal and/or evaporation-deposition processes.But the rate for grooving was the same in his experiments.

To have a better wetting process samples are sometimes heated in the liquid metalfor a short time to a higher temperature. This was done in our experiments occa­sionally. Such a process has to be selected very carefully. Areaction may startdifferent from that at experimental conditions.

3.2 Dissolution

The next step after wetting depends on several parameters. In most cases a thinlayer of liquid on the surface will be quickly saturated with the dissolving metal.Dissolved atoms have to diffuse through this layer into the bulk Pb-17Li. The con­centration in the molten metal increases until finally the bulk material is satura­ted and dissolution stops. On the other hand dissolution can be controlled by thedi sso1uti on of the metal. that means by transport of an atom from the soli d i ntothe liquid. In this case saturation needs much longer. the dissolution rate remainsapproximately constant for a longer time. However. a constant dissolution rate isalso seen in case of mass transfer. In our experiments it was isothermal masstransfer between dissimilar metals. in case of loops often non-isothermal masstransfer. In both cases dissolved atoms are removed from the liquid.

The equilibration time until saturation depends on the conditions of experiments.Solubility measurements are often done with equipments similar to those used in theexperiments of this report (6). The liquid metal is stirred to reduce the equili­bration time. Stevenson (21) investigated the dissolution of Cu and Ni in lead at527°C. The dissolution time constant increased by a factor of four when increasingthe liquid metal velocity at the sample surface from 8 to 20 cm/s. 4 Extrapolatingthe function would give equilibrium in a static system after 2 hours. On the otherhand, Ali Khan (22) found for the dissolution of iron in stirred lead equilibriumafter 50 hours at 1000 oe. This is a long time compared with Stevensons results.The exposure time in our experiments was always long. We assume, therefore, satura­tion in most experiments.

Solubilities of metals are similar in different molten metals (10). Even in lithiumsolubilities are often reported similar to those in lead or Pb-17Li. Theoreticalcalculations depend very strongly on the used model. For example. V solubilities inlead calculated by different models scatter over 8 orders of magnitude (10) !Therefore. casual similarities should not be used for predicting solubilities. Suchtheoretical considerations were not included in the discussions.

4 A different rneehanisrn was found with veloeities larger than 50 ern/s. This will not bediseussed in this paper.

y Guminski 's model looks promlslng (1e). It considers the electronic gura-tion of atoms in a 'cellular' model und finds similarities between different sol­vents. For example, Nb has a Ieee times lower solubility than Al in molten Li, Sn,Pb and Bi. This was shown by drawing solubilities oS a function of atomic numbers.Gur results will be compared with this model.

3.3 Hass transfer, diffusion coefficients

With most of the investigated metals, dissolution did not stop despite solubilitylimits were reached. Because of isothermal mass transfer between dissimilar metals(15) the dissolved metal deposited at the crucible wall. Different processes canreduce the chemical activity of a metal at the Mo wall, e.g.diffusion into Mo orthe formation of compounds. In some cases, however, the process can not be explai­ned.

The concentrat ion gradi ent across the cruci b1e remai ns constant and the Fi rstFick's law is valid. The Pb-l7Li is saturated at the sample surface, the chemicalactivity at the deposition surface is lower. The constant dissolution rate is givenby

R = D/x (Cs-Co).

An estimation of Co, the concentration at the crucible surface, is not possible.But assuming Co«Cs and the distance sample-crucible wall for x, diffusion coef­ficients can be calculated. However, one has to keep in mind that Co is not deter­mined and values should be considered with care.

3.4 Surface reaction zones

Different kinds of surface reaction zones can be formed : Penetration of Pb and Liinto the sample metal, formation of chemical compounds or leaching of elements froman alloy. The formation of a Ni-depleted ferritic layer on austenitic steels is anexample for the latter effect.

Chemical reaction zone

A new kind of a 'chemical reaction zone' was found. Reaction zones are usuallyi denti fi ed by metallograph i c exami nat ions. But often meta11 ograph i c exami nat ion sshowed no effect. However, the chemical behaviour during cleaning processes of me­tal s after exposure was di fferent Trom Hat of the ori gi nal metal: ~1ore of thesample metal was dissolved. We assurne that there was a more reactive surface reac­tion zone. The thickness of this was proportional to the exposure time. We believethat this zone is the first step before the 'metallographieal' reaction zone isformed and indicates e.g. penetration of lead along gain boundaries (s.steel 4922and other metals).

3.5 Intermetallic compounds

The formation of intermetallic compounds may occure with lead or lithium in solu­tion or at the surface. Depending on solubilities and transport properties, com­pounds may form surface reaction zones or deposit at the sample or crucible surfa­ce. All these effects were seen in the experiments for different metals.

/

1

3.6 Mutual solubilities

The corrosion continues even if the dissolution of a metal stops. Dissolution is amutual process. At the same time with dissolution in the eutectic, lithium and leadwill dissolve into the solid metal. This process is often accompanied by modifiedmechanical properties (50) or embrittlement, as seen for many metals. A typicalexample is beryllium in our experiments. Liquid mehl embrittlement is importantwith all liquid metals. Kamdar (81) gives a detailed description of the effect. Hedescribes that embrittlement was often found even if there was no measurable pene­trati on of the 1i qui d meta1 i nto the soli d one. That means that trace amounts ofthe liquid metal in the solid, e.g. dissolved at grain boundaries, are sufficient.

3.7 Gas phase reactions

The mentioned observation with grooving (19,20) leads to the question for reactionsin the gas phase. Indeed embrittlement of steels in a lead atmosphere is reportedby Kamdar (81).

The isothermal atmosphere in the capsule above the crucibles is nearly saturatedwith Li and Pb vapor 5, corresponding to the chemical activity of these elements.The upper section of a sample was exposed to this covergas. Reactions between thesample metal in this area and lead was observed with several metals, often causingembrittlement.

3.8 Complex mechanisms

Actually the dissolution and corrosion process is complex and all described me­chanisms will occure in parallel. Furthermore not a simple binary system has to beconsidered. Solid phases are the sample, the crucible wall, and formed intermetal­lic compounds or deposits. In the liquid phase at least two components, Pb andLiPb, are mixed. Reaction zones or leaching of alloy elements change chemical pro­perties of surfaces.

3.9 Some remarks on molybdenum crucibles

Crucibles of molybdenum were used in most of the experiments because of good compa­tibility of this metal with the molten eutectic. But with Mo, an additional solidmetal phase was introduced, dissimilar-metals mass transfer occured. Mutual inter­di ffusi on with Mo up to the formati on of compounds i s reported for many metals(25). The chemical activity of these metals at the Mo surface is reduced. Thequestion, if there is still a Mo crucible after deposition of e.g. Fe or V, can beanswered with 'yes'. Never any stop of the mass transfer process was observed.

At the beginning of the experimental program crucibles failed when a sample metalwas in contact with the Mo-wall below the liquid metal surface. This was observedin our experiments with Ti, Zr, Fe, steels and vanadium at temperatures above 450°C. Obviously the mobility of metal atoms is high enough at this temperatures formutual interdiffusion and chemical reactions. There was no systematic investigationof this effect; instead samples were fixed in the crucible to avoid this contact.But the effect has to be considered in connection with Mo coatings.

5 Not exactly saturation because of a very slow covergas flow.

4. Results ond discussion

All results are described in detail in Part-II of this report. together with adiscussion of the literature. In this chapter a kind of extended summary is given.Not for every metal all investigations were done. Also some informations may belost because of the long delay between experiments and evaluation. The readershoul d be aware that no menti on of an effect can mean no remarks were found inlaboratory notes. Nevertheless, a large number of valuable results were obtained.

4.1 Wetting by Pb-l1li

There was no systematic investigation of wetting behaviour, but some observationswere registered in laboratory notes.

A wetted surface after removing from the molten eutectic at 35e to 4ee oe retains afilm of 3e to lee microns of liquid metal, equivalent to 3e to lee mg/cm2 • Non-wet­ted surfaces are clean with droplets.

Wetting at 3ee oe was seen only w1th yttrium after 66 hours exposure time andw1th uranium after more than 35e hours. Also at 4ee oe, a longer time was oftenneeded for wetting. The wetting behaviour of different materials including Mo ofthe crucible was different. In agreement with Robertson's observations (19,20)however, even in case of non-wetted samples, solubility and dissolution rate datafit wel1 into the Arrhenius functions fram higher temperature experiments.

This important observation is in agreement w1th our experiments for gettering ofhydrogen (65) w1th vanadi um : wett i ng had no i nfluence on the rate of hydrogenuptake. The rate determi ni ng step was di ffusi on in a 1i qui d metal boundary 1ayer.This diffusion is probably dominant also in cases where wetting has no influence oncorrosion or dissolution, as seen for several metals.

Therefore. 'incubation times' as seen by Borgstedt (17) should be discussed betteron the basis of protecting surface layers than with wetting effects.

4.2 Dissolution and mass transfer

Oi ssolut i on rates were obtai ned from the sum of di sso1ved meta1 found in Pb-17l iand at the crucible wall. This corresponds to weight loss data which were determi­ned occasionally6. In case of Ti and Zr, reaction zones were growing at the surfa­ce, 'loss of sound metal' was higher than calculated from the dissolved amount.

6 Weight lass data were always in agreement with data from chemistry.

Ul

Only in a few cases an incubation time was observed before dissolution started. Adelay of 1200 hours was seen with uranium at 300 oe. But then dissolution wasfast. But even in thi s case it was not a wetti ng effect. An i ncubati on ti me wasseen wi th Ti and Zr metal (60), probably caused by remai ni ng surface oxi des. Exactinformations for these elements are lost. An other example for an incubation timewas with molybdenum coatings on Ti and Zr. Diffusion of Ti respectively Zr and Pbthrough the Mo layer eaused the delay. Then dissolution rates were the same aswith uneoated samples. 7

In most eases a faster dissolution during the first few hundred hours was followedby a steady state dissolution rate. Usually dissolution did not stop at saturationof the euteetie beeause of isothermal mass transfer to the erueible wall (below).Fig.l shows the dissolution of iron and steel 4922.

A faster dissolution rate at the beginning was found e.g. by Chopra (74) for ferri­tie steels in non-isothermal systems. 8eeause the initial period was not investiga­ted in this work, only steady state dissolution rates will be discussed. As mentio­ned in chapter 3.2, a constant dissolution rate is eaused by mass transfer.

All steady state dissolution rates as found in this work are shown in Fig.2, fune­tions are summarized in Table 1. For Mo, upper limits were found, for Y only alower limit at one temperature. No values ean be given for Nb, Ta and W.

Tab.1

Steady state dissolution ratesR in g/m2*d, T in K; heat of reaetion in kJ/mol.

ln R = A + 81T heat ofreaetion

A I 8

Al 13.6 -9560 -80Be 6.89 -9300 -77Fe 3.38 -7206 -60Ti 11.6 -9260 -77U 23.6 -12970 -108V 2.37 -5431 -45--,.~ 1 1 --, ooco -74LI

11

~~.I

1

-00;;0

11

-------- ------- ------ ------4922 4.06 -5195 -43

7 The pumped loop TRITEX (38) is constructed from this steel, the composition is similar tosteel HT-9 or MANET.

11

Fig.lDissolution of iron and steel 4922

in Pb-17Li4.0 -.---------------------.

.. 4922,,/ 550 oC,.

*"", •-,. .,. ,

'" '.,;,'" :'" ,'" ,,. ,

'" . '2.0- ......• -..- ~ ; ..., :

I

2000honrs

" : Y non,1 : ---T 600 oC

1 ~ ...... : .._,' __-Jt""'1 ..........

1 T"'" ...1 1

""11

O_&••o 4000

12

Fig.2Steady state dissolution rates

800 600 400 oe 300

0.001

, ," +

: Zr: : l1'< Y10- ; -:~ - - : - -- - : - - .. - -:- -'-, ~ -

Ti ~~,~:" :",. ........ ", . ,, ,." , . ... ..

1 ; - _.. -~~-~.,- ."':~ -_.. _.. - -;- - _ _. --: : ,,~~, .......... Al :, • .,....." I

: : ~.' ...... :, ....... ., '\0

0.1 - _.... . ... ; .... - - - .":' -~ -...; ~. -- - . -; .. -- -- .. -- ; -- - ....: '..., ." . .: ~,: :'" "'steel 4922: - '~ : .0.01- .. . . . . . ~ - _. '''-.-' . ~'''-; '"",. - ~ .. - - - _. _. - ~ - - _. - ..., -"""- ...... '~ ,

: .. : -~ ..... ~ '''''-.,~ V, i ' ~ ... ,.~ _. TMo' -_ ~~- Be- ':' _ .

: Fe

1.801.601.20 1.40

lOOO/T1.00

0.0001+----..,..,--_,---,r----"j"--,-----l.80

All functions have nearly the same slope, pointing to the same mechanism for disso­lution. Even in case of formation of compounds ,U and Zr) this slope is found.From the work of Stevenson (21) it can be concluded that diffusion in the liquidphase is the rate controlling step.

For the maximum blanket temperature of 450 GC (1), dissolution rates are shown inFig.3 . Three groups of metals can be identified. Most stable are the refractorymetals Nb, Ta, Mo, Re and W. The structura1 materials for a b1anket be10ng to thesecond group. From this group, beryllium cannot be used in Pb-17Li because of se­vere attack a10ng grain boundaries. Not usable in Pb-17Li are the elements of thethird group because of their high dissolution rates.

Corrosion rates in non-static loop systems are higher than in static tests. Thiscan be seen e.g. in Part-lI, Fig.4922-2. Also the slope of the temperature func­tion is steeper. Increasing the temperature from 400 to 450 GC causes a doub1ing ofthe steady state dissolution rate from our static experiments. The same temperaturerise wou1d give ten times higher corrosion rates for stee1s in loop systems(l7).Neverthe1ess at least a comparison of metals whith each other can be done. Figure 3can be used for the prediction of corrosion rates in other systems.

Isothermal mass transfer

Mass transfer from a samp1e to the crucib1e wall was seen with many of the investi­gated metals. The effect is described in the literature (e.g.ref.15). The strongestdeposit i on was seen wi th Al and compounds of U and Y. Be, Fe and V show al sostrong mass transfer, while probably the formation of compounds at the sample sur­face reduces mass transfer for Ti and Zr.

Mass transfer between dissimilar metals can p1ay an important ro1e for the inter­pretati on of corrosi on experi ments. For exampl e, Borgstedt found in 100p PI COLO(16) a weight gain of vanadium a110ys because of pick up of Fe from the eutectic.In our static isothermal experiments mass transfer caused a steady state dissolu­tion rate and gave the possibi1ity to ca1culate diffusion coefficients.

4.3 Solubilities of elements in Pb-l1li

The applied technique to investigate material behaviour in Pb-l7Li is often usedfor solubility measurements (6). After longer exposure time, the eutectic is satu­rated with the dissolving metal. Temperature functions for solubilities were obtai­ned for Al, Be, Fe, Mo, Ti, U, V and Zr; single -one temperature- va1ues for Y, Nband Ta; solubility limits for Re and W. Fig.4 shows all results. Functions andsingle values are summarized in Table 2.

14

Fig.3Steady state dissolution rates at 450 oe

100....--------------------.

0.0001....1.---

~[ID- - -INb, Ta, Mo, Re, W~ - -~~ - - - - - - - - - -0.01

0.001 _. - - - . - less than- ------ - -

00Q)~

e:tl~ 1 - -- - - . - - - . - . - - - - - - _ - - -- -- - - - - --~o.-~

:::::l-o~ 0.1

:.a

-""C...N

S 10""'-t:l.O--

15

Fig.4Solubility of elements in Pb-17Li

800 600 400 300 oe10000 ...........--I--+-~--+----I-----I-----.

1000

S 100

~

~

~ 10

0.1

• ... ... ... ......... Z ...... ~ .... r 0 :· -.......... y .· .... ....· " ... ... .,· '" .... U.'~ . ...· ~'" : ...

. . . . . . . . . . . . . . . .'. . ~.~ : ~ "' -..... ", .

............ : ~ ~~"'AI.... ....., .· ... .

...., : ............ Ti :....... : ...... '... :........ ... .", ... .

:~iY)ä..-(Re)"":~"Be· .•.............. : Fe... .......,;~

Ta v ...:......M? .~ V

NbO :0.01...1..-----,----;-------r-------t----,...----J

~ ""I.UU~ f") "I.LU

~ Ar.I . LtU

lOOO/T

• r" r.I.au 1.80

16

From these elements the solubility of Al in Pb-l7U was reported recentlyThe results are in good agreement. Only one value for the iron solubility at 450is reported (8). It is mud higher than values from our function. But this valuewas derived from stainless steel corrosion tests. so we found higher 'Fe-solubi­lities' in experiments with steels 4300 and 4922. This will be discussed be10w.

The solubilities of Ti, U, Y and probab1y Zr are actual1y solubi1ities of compoundswith lead (be10w).

Solubi1ities of a meta1 in different liquid metals are simi1ar. In fig.5 oursolubi1ity va1ues for 600 oe are given together with 'Guminski '5 function' (10) forlead. It is evident that the data fo110w the funetion.

An amazing fact is that also Ti, Zr, Y and U fit into the function. These fourelements form intermetallic eompounds with lead. Obvious1y solubilities are notmueh influeneed for these elements. In the figure a value for bismuth is inc1uded(31). Again in spite of the formation of Li 3Bi (32), the va1ue fits into the func­tion. However, when eonsidering the sea1e in Fig.S, the reader shou1d not expeet ahigh aceuracy for predieted solubi1ities from Guminski 's function.

Tab.2

Solubilities of elements.S in wppm; T in K; heat of dissolution in kJ/mo1.

1n S = A + BIT heat of singledisso- values

A B 1ution oe I wppm

Al 19.2 -10040 -84Be 14.2 -9446 -78Fe 4.94 -4292 -36Mo 10.0 -9784 -81Nb 600 0.OS3Re SOO < 1Ta 600 0.19ri 21. 3 -13600 -113U 2S.8 -121S0 -102V 10.1 -7730 -64W 600 <1Y 400 1100Zr 11.2 -3423 -28.S

1008040 60

Atomic Number20

Fig.5Solubility of elements in Pb-17Li at 600 oe,

compared with a function of Guminski (10) for leadlE+02 I " ,I: . / I: t: '.

" ,I. . ~ ~ " ' f' \lE+Ol- .:....:. i / ..\.... -:V\:-~··!·· .~.I. \. ~/\' .~....;-:.~...•...

lE+OO-:.! ~ .. \ +r·V;! {.. ~/\.. !Bi\i··lE-Ol- ·1·:· '.AI .. ~~ ~ I····· 'IZr':' ~ ~ ..;..: \/ .

'" , ·Tl I I ..' . I , ..Be! : I ~ ,: ..' : t . '.. ; .

I , : I t , ' : I 1 :.:! :~! ~ ·f· : ~ !..: .'v .I , ,

" ' " 1 , 1"tE-04- ·1········ .: .. Fe·"'" ·t'. . ..... ;. '(W Re'''')':-''''''''''

I' Mo., l"" ', , . e'lE-05- .. "'" :. . . . . . . . . . . . : ~ .

, Nb ' Ta'lE-06- -: ' .' ' .

:1-- Guminski. Pb( 10) ,e this work',.. Borker (8)'

1E-07+----:=,======::::;:'=======::;,;:::::======:;:=.,----lo

~ lE-02-.+.lro lE-03-

4.4 Diffusion cients in Pb-17

:1.8

Comparing Fig.2 and Fig.4 it can be seen that metals with high solubilities havealso high dissolution rates. From the First Fick's Law and assumptions as given inchapter 3.3, a linear relation is expected between steady state dissolution ratesand solubilities:

R = Dlx (Cs-Co).

Because of similar diffusion coefficients for different metals, this is found overthe range of five orders of magnitude, Fig.6 .

For 7 elements dissolution rates and solubilities were determined at the sametemperatures. For these, diffusion coefficients were calculated, using x = 1.2 cm(0.012 m) for the average distance sample-crucible wall. The values are in an ex­pected range (Fig.7). The functions are summarized in Table 3.

The elements U, Ti and Zr form intermetallic compounds at the surface. These com­pounds are in equilibrium with the solution. Fick's Law should be valid for com­pounds as for a pure metal and the data fit into the slope of Fig.6 . Diffusioncoefficients, however, may be influenced by these compounds.

The functions for Al, Be, Fe and V are considered to be more reliable for theseelements because no compounds are formed.

Included in the figure is a function of Robertson (19) for iron in lead. Inspite ofcompletely different experiments (study of 'grain boundary grooving') it has thesame slope as our Fe function, but the values are 10 times higher than ours. Aniron value of Simon (27) is obviously to low, while a chromium value of the sameauthor fits into the range of our functions. The low values of beryllium are pro­bably influenced by the odd behaviour of this metal : relative low solubility anddissolution rate, strong grain boundary attack. Not explainable is the positivevalue of the diffusion energy for vanadium.

Table 3

Diffusion coefficientso in m2/s, T in K; heat of diffusion in kJ/mol.

ln 0 = A + BIT heat ofdi ffus.

A B

Al -21.6 -920 - 7.7Be -25.4 171 1.5Fe -19.6 -2844 -24V -25.7 2300 19

----------------------------------------Ti -27.7 4348 36U -32.1 6302 52Zr -17.6 -5220 -43

Fig.6Dissolution rate vs. solubility at 450 oe

+4 y--------;------------:------,

+2 .... __ .

b.Do--2

- . - . . '. - .. - . - - - - - - - -," . - - . - .... - - - . - - .... - - - - - .., ,, ,, ,, ,, ,, ,, .. .: Al :, ,, ,. -----------': Ti' . _ , 0" .. , --_. --_..: 0 : Zr

+6+2 +4log 10 S(wppm)

o- 4+.-----,---,r---r---;---...,.---;----,----f

-2

28

Fig.7Diffusion coefficients of metals in Pb-17Li

700 600 500 400 oe1E-8 ...,.-I:I----t--:---I:I------I:----:-------,

.... ... ... :Robertson (19)...... ... ,Fe in Pb· :

1E- 9 - " ":' ;' . , , , . , , , , , , .. :. : .-. : ,@] .. ,: ,~: ... ,

00 . -:., ~ : .~ , .. ' :......-.......- :1o'L ••• ~. .......-~ •••• ,S ......-..."..".",. .; rA'il.~ _...~~~.':'~' ~ ~ 'IF~r""" ~'~'-'~ ... ,... "",

.. ':. . • Simon (27) :, '. er in Pb-17Li '

".1E- 11 .. . ~,~ ~ ,1· ,~ ,~ ,,~ ,~ IB~ r' .. :..~ .

Simo~ (27).. Fe in Pb-17Li

1E-12 D=6.0E-t4I I I

1.00 1.20 1.40 1.601000/T

1.80

4.5 Reaction Iones and compounds

Some reaetion lones and eompounds were mentioned in the ehapters before. Reaetionlones at sample surfaees may be formed by different proeesses. The ehemieal aetiv­ity of an element in a reaeti on lone or eompound i s sma11 er than for the puremetal. Therefore. the eompatibility with the euteetie will be influeneed. Differentkinds of reaetion lones were found in this work.

4.5.1 'Chemical reaction zone'

Chemieal purifieation of samples showed for most investigated metals very thinreaetion lones. The thiekness was proportional to exposure times and larger athigher temperatures. Typieally on vanadium the thiekness was e.e4 mikrons afterexposure to the euteetie for 3eee hours at 55e eC. These reaetion lones ean not beseen by metallographie examination. We assume that such ehemieal reaetion lonespoint to the beginning of dissolution of lead into the metal. For example. thestart of a grain boundary attaek on iron was seen at 6ee eC after 156e hours. whilethe ehemieal reaetion zone was seen also at lower temperatures and shorter exposuretimes.

4.5.2 Metallographical reaction zone

Metallographieal reaetion zones are most often reported in the literature. Für puremetals and without formation of eompounds they are eaused by diffusion of solventatoms into a sample metal. Mainly lead is penetrating along grain boundaries. Sam­ples get brittle and ean be destroyed, as seen with beryllium. Grain boundaryattack was seen also with Fe, Mo and other metals.

4.5.3 Chemical compounds on surfaces

A different kind of surfaee layer is seen when eompounds are formed.

Metals may pick up elements oS 0, N or C fram the eutectic cr atmosphere. This wasseen with vanadium. But the reaetion zone was thin and did not prevent the dissolu­tion of this metal.

Much more important are reaeti ons wi th 1ead from the euteet i c. In alleases theformation of eompounds eaused destruetion of the sol id metal. The following eom­pounds were found :

titaniumuraniumyttri umzireonium

TizPb and Ti 3PbzUPb q

YPbq

ZrqPb

From these eompounds Ti 3Pbz• YPbq and UPb'l were not reported before. TizPb wasproposed by Farrar (62) for the Ti-Pb system. but not experimentally verified.

22

Ti 3Pb 2 was investigated more in detail (23). Well shaped crystals were found in theexperiments. The crystals are chemically much more stable than Ti metal. The compo­sition of the crystals gives Ti-jJb z or a more complex compound. In the literatureno compounds are reported with more than 33 at.% Pb. From some selected crystalscrystal10graphic parameters were determined : space group P63/mmm, with a=0.93 andc=0.58 nm spacing.

In thi s new compound Pb and/or Ti can be rep1aced by other metals. There was nosystematic investigation of this effect. Other elements were found only when usingTi-a110ys as discussed below. The crystals contained in our experiments the fol10­wing elements:

Zr up to 14 at.%Al 40.6 at.%V 0.27 at.%Mo 0.27 at.%Sn 0.63 at.%

A partly replacement of Ti by Zr over the whole concentration range was found e.g.for the intermetallic compound Ti/ZrsPb 3 (63,64). A very high concentration of Al,however, shows that areplacement of lead is also possib1e. More tests are neededto quantify these effects.

The dissolved titanium is obviously in equilibrium with the intermetallic compoundTi 2Pb. The same concentration in the eutectic was found with Ti metal and with allinvestigated alloys. Also zirconium is forming an intermetallic compound with leadat the surface. The equilibrium concentration, however, is proportional to X(Zr) inthe alloy.

4.5.4 Chemical compounds in solution

The compound Ti 3Pbz is formed by crystallizing from the solution. Therefore, thereplacement of Ti or Pb by other metals is possible. As in aqueous systems, deposi­tion occurs if the concentration exceeds the solubility product. Depending on solu­bil i ti es and the effect of mass transfer, depositi on at the sampl e or on othersurfaces is possible. Compounds with lead were found also at the crucible wall.

23

4.6 Behaviour of al10ys

Results with alloys have been mentioned in the chapters before. The behaviour ofalloys is often different from that of pure metals, even when considering reducedchemical activities of the components. All investigated alloys showed no betterbehaviour than the pure base metals.

Alloys of Ti and Zr were studied as possible getter metals for tritium. The concen­tration of Ti in Pb-l7li with these alloys showed that the compound TizPb, formedas a surface layer, determines the solubility of Ti : independent on X(Ti) in thealloy, the same value was found for the equilibrium concentration. The compound ofthe crystals, Ti 3Pb z has obviously no influence. On the other hand, equilibriumconcentrati ons of zi rconi um in Pb-17li are di rectly proporti onal to X(Zr) in thealloy, although the surface reaction product IrqPb was formed. But the concentra­tion for X(Zr)=l, derived from alloys, is lower than for pure Zr metal. That meansthat the chemical activity of Zr in alloys is lower than X(Zr).

Titanium and zirconium were leached from the Mo alloy TZM. He leaching rate issmall. This alloy is as stable as Mo metal.

A few alloys with aluminium showed that this element is leached out from Ti-V-Aland Ti-Zr-Al. The Al-concentration in the eutectic was very low compared to thesolubility, clearly indicating the formation of compounds. Nevertheless, crystalsof pure Al were found at the crucible wall. The chemistry of aluminium should beinvestigated more in detail because of its importance for alumina coatings (30).

Austenitic steels showed the typical ferritic surface layer because of Ni-lea­ching. More important for a blanket is the behaviour of the ferritic steels 4922and 4301. The dissolution rates are higher than for pure iron. Also the equilibriumconcentrations of Fe in Pb-17li are higher than the Fe-solubility ! At 450 Ge wefind for pure iron a solubility of 0.4 wppm, for the steels an equilibrium concen­tration of 0.8 wppm. Barker (8) found for the austenitic steel 316 an equilibriumvalue as high as 20 wppm. Probably not pure Fe is dissolved but a compound. On theother hand, ni cke1 with on1y 0.3% instee1 4922 showed i dea1 behavi our. It wasleached out even fram ferritic steels. This leaching may have caused the chemicalreaction zone and the start of an intercrystaline attack. For equilibrium concen­trations of chromium data were scattered as usually (6,8). For manganese, the ac­tivity coefficient from equilibrium concentrations is in the range of 0.1. Butthere may be an other explanation for the 'too low' Mn values. Even if not found instatic compatibility tests, the compound MnNi may be formed at the surface, with amud lower solubility. In a non-isothermal stainless steel convection loop MnNiwas found at the coldest spot (33,34). This compound is proposed for the removal ofMn-54 from the eutectic in a blanket (35).

Summarizing the behaviour of alloys we have to conclude that alloys and the role ofcompounds in solution should be studied much more in detail in order to understandtheir behaviour in the eutectic. So far, not all effects can be explained.

24

4.7 Mo coatings

When sample metals contacted the molybdenum of the crucibles below the liquid metalsurface, failures of crucibles were observed. It was mentioned in chapter 3 thatinter-diffusion of solid metals or even chemical reactions are responsible for thiseffect (25). In the literature it was reported that Mo coating on steels was notstable (36,37). Nevertheless, Mo coatings on getter metals were investigated.

The coatings showed a very limited protection effect for Ti, Zr, Y and alloyBeta-3. In case of titanium, for example, Ti as well as Pb are diffusing through Moat 500 oe. The described reaction layer of Ti 2Pb on Ti metal is seen below Mo. Alsocrystals of Ti 3Pb 2 are formed on the surface of Mo. The incubation time wasdiscussed before in chapter 4.2.

4.8 Solubilities of Li and Pb in solid metals

The solubility of metals is a mutual process. Therefore Pb and Li will dissolve inthe solid sample. Several attempts were made to determine Li and Pb in the samplemetals.

lithium: Li was never detected because of a too low sensitivity of the analyticalmethod. The concentration in all materials was below 0.1 wt.%. However, Li is cle­arly penetrating into the metals. Steel 4922 samples from loop TRITEX were chemi­cally cleaned, even a surface layer was removed. After some time in air, Li isdiffusing from the metal to the surface, forming there alkaline reaction products.This is a well known effect from all sodium systems. Coen (28) found Li in austeni­tic steels always together with lead. He assumes areaction of Li with oxygen ornitrogen in the steel. But intergranular penetration without such reactions is alsopossible as described e.g. by Wilkinson (29) for steel 21/4Cr-lMo.

lead: Also for Pb results are meager. Boundary concentrations e.g. for Al ( <500wppm), V ( <200 wppm) or Be «1000 wppm) are of limited value. Only for titaniumthe solubility for lead in solid Ti was found. It is given by

ln S(wt.%) 19.9 - 15980/T

The results add new data points to the phase diagram (23), (Fig.8).

25

Fig.8Phase diagram Ti-Pb fram ref.(61)

1000.------------------,

•••••••••••.-.....- ----JII'...-_--I

•.•..

I<> this work I<>

<>.8

uoCl)

8800-+-:lco~Cl)

~

SCl)

E-600

400~----r-1-----r--j--....,..---.,....,-

o 2 4 6 8at.% Pb in Ti

4.9 Gas phase reactions

Inspite of its importance for embrittlement no systematic investigation of the gasphase part of samples was done.

As mentioned in chapter 3 reactions between lead and U or Y were seen in that partof the sample which was exposed only to the gas phase above the eutectic. Withthese two metals even a kind of 'super wetting' was seen : 'creeping up' of theliquid into the covergas range of the sample. The mobility of lead in solid U andY, therefore, is much higher than in other metals.

Embrittlement of the gas phase part of a sample was seen with some other metals.The effect was not investigated in detail. Only aluminium and beryllium should bementioned in this chapter. Heavy embrittlement was seen in all parts of the sample.

21

5. Sunmory

The compatibility of 31 metals, alloys emd coatings with static eutectic mixturePb-17Li was investigated in more than 300 tests. Most of the results have not beenpublished before.

No influence of the Li concentration on results was seen between 0.62 and 0.77wt.%. Wetting has no influence on dissolution rates. This is discussed in detail.Metals can be divided into three groups : Most stable are the refractories Nb, Ta,Mo, Re and W. Ferritic steels, Be, Fe, and V belong to the next group. However, Beis destroyed along grain boundaries. Not stable at all are Al, Ti, Zr, Y, U andits alloys.

Temperature functions for solubilities in Pb-17U were obtained for 8 elements,single -one temperature- values for 3 others. The results are in good agreementwith a theoretical work of Guminski. Remarkably high are solubilities of Al. Zr. Yand U. while those of the refractories are low. Also, the solubility of Pb in solidTi was determined, adding new data points to the phase diagram.

Because of the effect of mass transfer between dissimilar metals diffusion coeffi­eients in Pb-17Li could be calculated from dissolution rates and solubilities. Mostreliable are the temperature functions for Be, Al. Fe. and V. Those for Ti. Zr, Yand U are influenced by the formation of compounds. All values are in an expectedrange, but not all effects can be explained.

Different kinds of reaction Iones were found on surfaces. New is a very thin 'che­mical reaction zone', identified for several metals during sample cleaning. It isprobably formed as a first step before grain boundary attack of the eutectic.

The following new intermetallic compounds were identified: Ti 2Pb and Ti 3Pb 2 • UPb4 ,

YPb 4 and Zr4Pb. The compound Pb2Ti 3 was i nvesti gated in detail. Lead and titani umcan be replaced by other metals. With Y and U, there was even areaction in the gasphase above the eutectic.

Generally, alloys were not more stable than their base metals. Leaching of ele­ments from a110ys and other effects were investigated. Especially with alloys. manyquestions remain open and more work has to be done to understand the chemistry ofa110ys in the eutectic.

Last but not least Mo coatings on getter metals were found not to be protective forthe use in a blanket.

29

Part-lI : Results

AluminiumBeryll iumIronMolybdenumNiobium, TantalumRheniumTitaniumUran i umVanadiumTungstenYttriumZirconium

Ti-Zr - alloysalloy Beta-3Ti-V-Al - alloysTi-Zr-Al (SAES-getter)Mo-Re all oys,alloy TZM,austenitic stainless steelsferritic steel 4922

Coatings of Mo on Y, Ti, Beta-3 and Zr

313941556161698185939599

101111113111121121121122

121

41

alpha-Ircm==========

Only a few experiments are reported for pure iron in lead or Pb-17li. But iron isthe mai n component of stee1sand many corrosi on experi ments have been performedwith steels. Barker (8) remarked that iron in ferritic steels should behave simi­1ar to pure i ron. Not much i s known about phase di agrams of Fe with Pb or Li.Mutual solubilities are low even at temperatures above the melting point of iron,no intermetallic compounds are reported (12,49).

1. Experiments

The table shows the performed experiments. Iron was obtained from Goodfellow, with99.5 % quality. It contained 0.3 % Mn, 0.1 % Si, 0.08% C, 0.04 % P and 0.05 % S.The metal was cleaned and vacuum annealed for several hours at 600°C.

2. Results

There were no remarks about wetti ng of i ron in our 1aboratory notes. Because ofhigh temperatures of the experiments we assume wetting in all experiments.

Dissolution

Fig.Fe-l shows the results of all experiments. Because of mass transfer to thecrucible wall, the dissolution does not stop after the initial period. Steady statedissolution rates. Fig.Fe-2 , are given by

ln R = 3.38 - 7206/T

(R in g/m2*d). The heat of activation is -60 kJ/mol.Of course, dissolution rates from flowing Pb-17U as obtained e.g. by Borgstedt

(50) or others for ferritic steels, are up to three orders of magnitude higher.But such dissolution rates should not be compared with each other. The dissolutionmechanism is different, as can be seen from the high activation energy of -200kJ/mol found by Borgstedt.

Solubility

Two solubility values were obtained in this work (Fig.Fe-3 ). From these the solu­bility is given by :

ln S( wppm) 4.94 - 4292/T

The heat of dissolution is -36 kJ/mol.

A solubility function from Coen (51) gives obviously too high values. Also a valueof Barker (8) is much higher than solubilities found in this paper. Barker's valuewas deri ved from experi ments with stee1 316. We found also in experi ments withsteels higher 'solubility' values. They are inlcuded in Fig.Fe-3 and discussed be­low together with compatibility of steels.

48

toare not

than

ilead

time5 hito ours.

Stevenson found for the 501 ub1 11 1n at 600 oe a val ueours, but s fl.lnet1on 1s steeper • Al 1 Khan' s funet 1ons ,cl understandable. Data taken from fig.13 of ref.(S3) are 10from our function. The function from Weeks (54) 1s nearly identi

Two diffusion coefficients were calculated trom values tor steady state dissolu­tion rates and solubilities :

D (See 0C)D (6ee 0C)

7.48*le-11 (m2/s)1.14*1(:)-10 (m2/s)

The Arrhenius funetion is given by

ln D(m2/s) = -19.64 - 2844/T

the activation energy for the diffusion proeess is -24 kJ/mol. Fig.Fe-4 shows thefunction.The values are in an expeeted range. The extremely low value of Simon (27) for Fe

in the euteetie with 6.e*1(:)-14 m2/s (See 0C) eannot be explained. The same authorfound for chromium a value of 4.e*11-11 m2/s, in the same range as the iron value inour work. Robertson (19) measured diffusi on eoeffi ci ents of i ron in 1ead. Hi svalues are le times higher than found in this work.

Mass transfer. reaction zone

There are a number of intermetallie eompounds in the system Fe-Mo (49). No eom­pounds eould be identified by ehemieal analysis or metallographie examination ofthe crucible wall. Mass transfer was observed before in lead systems (15.36). alsowithout identification of compounds. No reaetion products could be seen at the ironside. But a thin ehemieal reaetion zone as deseribed for other elements points toa modifi ed surfaee. An even di sso1uti on at the surfaee at see 0C ean be seen inFig.Fe-5 . At 6ee sC. Fig.fe-Ei, a grain boundary attack starts. Ali Khan (22.53)reports sueh a reaetion between iron and lead. but for mueh higher temperatures.

3. Summary

Alpha-iron is relatively stable in Pb-17Li up to 6ee 0C. At this temperaturegrain boundaries are attacked. The solubility of alpha-iron in Pb-17Li was determi­ned at see and 6ee 0C. The values are mueh lower than data from Barker or Coen, butin the range of values for lead. Also. from steady state dissolution rates, diffu­sion eoeffieients were caleulated. These values are Ieee times higher than avalue of Simon.

49

List of experiments with iran.Header line temperatures oe, other numbers hoursexposure time.

-I 500 I 600 r-----------------

380 3600960

1152 108014401560

1776 167518002110 2180

26842690

-----------------

58

Fig. 1Dissolution of iron in Pb-17Li

2.0.,...--------:----------,

•.'•••••••.'•....:..•.... :1.0 ..............•.............................

••••••

4000

.600oC

'Y 500 oC

..,,

:' _Y- - - -~~: ""'.".. - - _~ I

' '''O-lo<---------t--------~

o 2000hours

Fig. Fe-2Steady state dissolution rate of iron

600 500 400 oe1E-11.,...----+--:--,---1--:-----.,.--+--:----.

1E-2- • • • • _ _ ...... • M • • • _ ~ • _ _ _ , _ _ • • _ • • • _ _ • • • _ _ • • • • _

... '

1:9... :..."l..... ... ...

"'51 ... ...

1E-} .....

1E A

1.00 1.20

1000/T1.40 1.60

51

Fig. Fe-3Solubility of iron in Pb-17Li

600 500 400 oC100.........----1--.-..,..--....,.1-------+.---..

o :Borker(8)

0.1 - - -

10 -_._._.~---:--------------~----_._-_.

S .-........:... Khon(22,53) I IQ., :8.............. Pb.this worke- ',. 4922 ................... ", <>

.~~~,~~oo ~ __ ... _

",~~, :· " .........· , '.. Weeks(54)· " : ",Pb,· , .

. . - - ; - - - - - - - - '-. - - - - - -. - - - - - . - - - .

,, , : Stev.(52)

:' Pb. ,. ,

1.601.40·1.200.01+-------------;---.:::...-----1

1.001000/T

Fig.Fe-4Diffusion coefficients of iron in Pb-17Li

700 600 500 400 oC1E-8....-+-.---+-:-..,..---1-.------1:-----.

1E-9

r--_ .-_ Robertson (19)

--_ Fe in Pb---~----

. .~:;~~:-------_.

+Simon (27)er in Pb-17Li

1E-11 - - - -

1.601.401.20

Simon (27)Fe in Pb-17Li

1E_12-t- -r-__·_D_=_6.0_E_-1,..4 -II

1.00

1000/T

S2

Fig. Fe-5

Fe metal after 1776 hours at see oe in Pb-17li.2 positions.(etching 2% HN03 + ethanol)

.-20 m

-

53

Fig.Fe-6

Fe metal after 1560 hours at 600 oe in Pb-17li.2 positions.(etching as before)

..

S5

Molybdenum===========

Because of its outstanding high temperature properties and its stability in moltenmetals. Mo may be considered as a structural material for liquid metal blankets. Acompilation of binary Mo-systems is given by Brewer (25). Molybdenum forms interme­tallic compounds with many elements and mutual inter-diffusion in the solid metalsi s reported . The i sotherma1 mass transfer as observed with many metals in ourexperiments may be connected to this effect.

1. Experiments

All molybdenum. samples as well as crucibles. were chemically polished and vacuumannealed at 700 GC for 2-4 hours to a final pressure of 10-5 mbar. Using Goodfellow99.9+% quality molybdenum in armco-iron crucibles. 11 experiments were performed.In addition. nearly all other tests were performed with Mo-crucibles. These cruci­bles were obtained from Plansee company. the purity was 99.95+ % . Therefore. ana­lysis for Mo was performed in many experiments.

2. Resul ts

Wetting was always good in experiments with temperatures of 500 GC and higher.Poor wetting was seen below 500 hours at 400 GC. no wetting in a 300 GC/66 hoursexperiment.

Dissolution 4nd solubility

The analytical method was not sensitive enough to detect the solubility of Mo inPb-17Li in experiments with iron crucibles. Also. Mo may have been covered by ironbecause of mass transfer (refer to chapter iron). Only upper limits can be givenfor dissolution rates: <2*10-3 g/m2*d at 600 GC and <4*10-3 g/m2*d at 645 GC.

With an improved technique, developed especially for Nb and Ta, values were obtai­ned. In four experiments crucibles were heated for 5280 hours to 600 GC. 0.02±0.005 mg Mo were dissolved. No Mo was found on the surface of Nb and Ta samples,all Mo was dissolved. Saturation can be assumed after such a long exposure time.The solubility of Mo in Pb-17Li at 600 GC is

0.3 ± 0.1 wppm.

Fig.Mo-l shows solubilities of Mo in Pb-17Li and lead. The function of Coen (51)for the solubility in the eutectic was a proposal but never published again. Thevalues are too high. For lead Shepard (55) gives 3 wppm for 816 GC, Brasunas (15)10 wppm at 1000 GC. It can be seen in the figure that these two values and our 0.3wppm for 600 GC in Pb-17Li fit very well an Arrhenius-slope, given by

ln S (wppm) = 10.0 - 9784/T

The heat of dissolution is -81 kJ/mol.A boundary value from Alden (56) for 1200 GC fits in this function, while a value

from Asher (36) for 700 Ge is 10 times lower than the function. A function compiledby Brewer (25) is orders of magnitude lower and not considered to be correct.

56

(The solubility of Mo in Li is reported by Leavenworth ( A wppm-function wouldfit well in Fig.Mo-l. Because of the large difference in the molecular weightsLi-eutectic/Pb, an appm-funetion gives mueh lower values.)

Reaction zone

It is not elear from the results if there is a ehemieal reaetion lone on Mo fromreaetions with the euteetie. Even if during eleaning proeesses more Mo was dissol­ved from erueible walls than from fresh Mo, results seem not to depend on tempera­ture or exposure time. Chemieal reaetion lones were found e.g. below deposited Tior Zr. In this ease, however, reaetions may have oecured with the deposited metals.

There was never any embrittlement seen on Mo samples or erucibles. But from metal­lographie pictures (Fig.Mo-2 ,md 3), an attack of the surface, as well as inter­crystalline penetration, is evident. The severity of the effect was a funetion ofti me emd temperature. Such attack was observed for hi gher temperatures in 1eadbefore. Only Grassi saw no attack after 500 h at 1093 0C in lead. This was probablya problem of preparation when removing lead. He found modified mechanical proper­ties, pointing to areaction (24).

3. Summary

As expeeted Molybdenum is very stable in the eutectic. Its solubility at 600°Ccould be measured, it is only 0.3 wppm. A good Arrhenius function was found toget­her with data for lead. Inter-crystalline penetration and a modified surfaee struc­ture was seen after exposure at higher temperature for a longer time.

Table: List of experiments with molybdenum.in iron crucibles.Header line temperatures 0C, other numbers hoursof exposure time.

550

610610

1308

48504850

600 645 r137354617

ll0018801880

With only a few exceptions, testsfor other metals were performed inMo erucibles. These tests were alsoexperiments for molybdenum.

51

Fig. Mo-lSolubility of Mo in Pb-17Li and Pb

1000 800 600 oe

!+thiS work IShe~hord (55)

Coen (51).------..:..., . -----

· . .· .Brosunos (15).: : _

1 .....

10

· .O. 1 ...........: . . . . Asher (36~. e ........ ~ ...........

0.6 0.8 1.01000/T

1.2

58

Mo metal after 5280 hours at 6BB oe in Pb-17Li.2 positions.

Fig.Mo-J

Mo metal after 1880 hours at 645 oe in Pb-17li.2 positi ons.

61

Niobium and tantalum====================

Metals like niobium or tantalum will be used only in special cases for fusionreactor blankets. Only a few tests were performed when looking for materials whichcould be used in experimental facilities. No phase diagrams were reported for Nband Ta with Li or Pb.

1. Experiments

The tab1es show the performed experi ments. Ni obi um was obtai ned from Goodfe11 owwith 99.9+ % , tantalum with 99.8+ % purity from Plansee company. The metals werecleaned and vacuum degassed at 800°C for several hours. The behaviour of Nb and Tais very sensitive against oxygen dissolved in the metals. Klueh (58) describes thepenetration of lithium along grain bounderies if the oxygen concentration is ashigh as Ie00 ppm. Rumbaut (68) mentioned preferential leaching of Nb from Nb-Moalloys in lithium, this may have been because of nitrogen in lithium.

We assume, however, no such effect in our experiments. The oxygen content in bothmetals was low, it was further decreased by the process of vacuum degassing at800°C. The solubility of oxygen and nitrogen in the eutectic is very low, no trans­port is expected.

2. Results

Good wett i ng was observed in all 500 and 600°C experi ments. Both elements areextremely stable in the eutectic. In most tests neither dissolution rates nor solu­bilities were obtained. The only values were from the longest experiment: 5280hours at 600°C. As averages from two crucibles only 6.2*10-4 mg/cm2 Nb and 2.2*10-3

mg/cm2 Ta were dissolved. Assuming these are values for saturation, the solubili­ties at 600°C are:

Nb 0.053 wppmTa 0.19 wppm

The values are shown in Fig.NbTa-1. But even to get these values, the extractortechnique for bulk Pb-17Li dissolution had to be modified : Pb was precipitated asPb(N03Jz and removed from the solutions. All values were near deteetion limits.There were a few older experiments with Nb and Ta in lead. The results are eompiledby Ali Khan (22). All tests were done at temperatures around 1000 0C. Even if theresults are not eonsistent, corrosion of both metals at this temperature was evi­dent. Ali Khans own measurements gave 0.001 ppm Ta in lead at 1000 0C, a value muehlower than our value at 600°C.

No mass transfer to the crueible wall was seen. No chemical reaction zone wasdetected (less than 0.01 microns). Metallographie examinations showed no attaek orreaction zone. Only at a few spots at a Nb sample, week intererystalline reactionwas seen. Fig.Nb-l, Fig. Ta-I.

62

3. Summary

Nb emd Ta are stable in the eutectic. Soll.lbilities at 600 oe could bedetermi • they are very low. No reaction zones were identified.

Table: st of experiments with niobium.Header line temperatures oe. other nl.lmbers hoursof exposl.lre time.

11381138

460 500 600 r137 192

644644

826

528052805280

Table: List of experiments with tantalum.Header line temperatl.lres oe. other nl.lmbers hOl.lrsof exposl.lre time.

1 460 500 600 r137 192

644644

826528652805280

63

Fig. Nb,Ta-lSolubility of Nb and Ta in Pb-17Li

1000 800 600 500 400 oe.

• : Ta• I • •... - - - .. - . - - - - - . - - - . - - - .. - - ..• • I •

• : Nb

0.01

1.601.20 1.401000/T

1.00

Khan (22)Ta in Pb

0.001.&...-._..6----.-----.-----.-----.-----'......80

64

Fig.Nb-l

Nb metal after 5280 hours at 600 oe in Pb-17li.2 positions.(etching glycerol/HF/HN03 1:1:1)

65

Fig.Ta-l

Nb metal after 5280 hours at 600 oe in Pb-17li.2 positions.(etching CiS Nb)

Titanium========

Titanium was proposed as a tritium getter and studied as one of the first metals(60). Most of the results have been published (23). For completeness, the resultsare given with more details in this report. The phase diagram Ti-Pb is investigatedin detail (12,61). A number of intermetallic compounds are identified, but not allcompounds were found by all authors.

1. Experiments

He table shows the performed experiments. Only experiments are included whichwere used for data evaluations. In addition, 6 experiments at 550 Ge were performedwith pure lead instead of Pb-17Li. The titanium was obtained from Goodfellow with99.6+% quality. The metal was chemically cleaned and degassed at 700 Ge for atleast 2 hours at a final pressure of 10-5 mbar.

2. Results

There was only a minor isothermal mass transfer observed to the crucible wall.Different to most other metals, however, the behaviour of Ti in lead or the eu tec­tic is controlled by the formation of intermetallic Ti-Pb compounds (23). Leadreacts with titanium, forming a surface layer of reaction products. After longertime at temperatures above 450 Ge well shaped crystals are growing on the Tisurface.

Dissolution

Steady state dissolution was seen after an 'incubcation time'. The weight of formedcrystals was added to find the 'total dissolved amount'. Unfortunately this weightwas often not determi ned and not too many data poi nts exi st. The formed reacti onzone is not included. He so defined steady state dissolution rate should not becompared with those for other metals. It is given by (23)

In R = 11.6 - 9260/T

(R in gjm2*d). The heat of activation for the process is -77 kJjmol. The func­tion is included in Fig.Zr-l. Data, however, are widely scattered, the function maybe considered only as an approximation.

79

Solubili of Ti in Pb-17li

Fig.Ti-l gives the solubi1ity of Ti in Pb-17Li. Actua11y the function correspondsto the solubility of Pb-Ti compounds. As discussed be10w with a110ys the chemica1activity of Ti in TizPb, the compound of the reaction zone, determines the solubi­1ity function found in this work.

The function is given by

1n S(wppm) 21.3 - 13600/T

The heat of dissolution is -113 kJ/mo1.

Weeks (54) has extrapol ated a val ue of 6 wppm Ti in 1ead for 500 Ge from Bi -Pbdata. This va1ue may be on1y a rough estimation, it was derived from on1y two datapoints with a solubi1ity of 1400 ppm for pure Bi. Our function wou1d give 41 wppm.Higher solubi1ities in Pb-17Li than in lead can often be expected as discussed inPart-I of this report. However, our lead va1ues for 550 Ge fits into the functionfor the eutectic. Neverthe1ess, our solubility function is assumed to be reliab1ebecause derived from many data points.

From solubilities and the steady state dissolution rates diffusion coefficientswere ca1cu1ated, Fig.Ti-2. He va1ues are in an expected range. The Arrheniusfunction is given by

1n 0 = -27.7 + 4348/T,

the heat of diffusion is +36 kJ/mo1.

Solubility of Pb in Ti

After removing reaction 1ayers from the surface the solubility of Pb in solid Ticou1d be determined by disso1ving the remaining metal. Fig.Ti-3 shows the resu1ts.The function is given by

1n S(wt.%) = 19.9 - 15980/T

(S in wt.% I!). He heat of dissolution is -133 kJ/mol. He solubilities arerather high, but fit we11 with a proposed 1ine of the phase diagram (61), Fig.Ti-4.There is a homogenious distribution of this disso1ved Pb over the cross section ofthe samp1e, as can be seen in the norma1ized microprobe picture, Fig.Ti-5.

11

Metallographic examination of samples shows reaction zones at the surface. Examplesare given in Fig.Ti-6 and Fig.Ti-7. The reaction zone is less stable than Ti metal.It can be dissolved in nitric acid. but it remains intact during the electrolyticpurification. The reaction zone grow thicker at higher temperature and with longerexposure time. but the maximum thickness found was 20 mikrons. Because of dissolu­tion and formation of crystals. the sound metal gets thinner and is finally de­stroyed. The same kind of reaction zone was found in experiments with Pb-17li aswell as with pure lead. Brasunas (15) found no reaction zone at 1000 GC in lead.Probably the dissolution rate of Ti or formed compounds is too high at this tempe­rature.

Chemical and microprobe analysis (Fig. Ti-5) shows that the reaction zone consistsof a Pb-Ti compound wi th homogeni ous compositi on : 65 ± 3 at.% Ti and 35 ±3 at.%Pb. We assume that Ti 2Pb i s formed. a compound proposed by Farrar (62) for theTi-Pb system. but identified for the first time in our work(23).

Formation of crystals

As was mentioned before. well shaped crystals were growing or deposited on the Tisurface. In some experiments crystals were also seen at the crucible wall. Thesecrystals remain at the surface during treatment with acetic acid/peroxid or elec­trolytic cleaning and are even stable in hot 2m nitric acid. They are also stablein fresh Pb-I7Li at 720 GC for more than one day. However. the crystals can easilybe scraped off. Fig.Ti-8 shows some of the crystals. The amount of formed crystalsas a function of temperature is given in Fig.Ti-9. The crystals have a homogeniouscomposition (Fig.Ti-5). Independent on temperature and exposure time. they containin experi ments with pure 1ead 56 at.% Ti and 44 at.% Pb - with a very narrowrange of only ± 1 at.% ! In experiments with Pb-17Li the composition was 57 at.%Ti. 42.6 at.% Pb and 0.4 at.% Li.

In the 1iterature no compounds are reported wi th more than 33 at.% Pb. The foundcomposition of the crystals would give Ti JPb2 or a more complex compound. With someselected crystals. crystallographic parameters were determined space groupP6 3/mmm. with a=0.93 and c=0.58 nm spacing.

In this new compound. Pb and/or Ti can be replaced by other metals. There was nosystematic investigation of this effect. Other elements were found only when usingTi-alloys as discussed below. The crystals contained in our experiments the follo­wing elements:

Zr up to 14 at.%Al 40.6 at.%V 8.27 at.%Mo 8.27 at.%Sn 8.63 at.%

A partly replacement of Ti by Zr over the whole concentration range was found e.g.for the intermetallic compound Ti/lrsPb 3 (63.64). The very high concentration of Al.however. points to areplacement of lead. More tests would be needed to quantifythese effects.

72

Experiments with pure lead

The rate of reaction as well as the formation rate of crystals were slightly smal­ler compared to experiments with the eutectic. All other results were identical, ascan be seen in the figures.

3. Summary

Except at lower temperatures Ti cannot be used to extract tritium from the moltenmixture. Because of reactions between the metals, the compound Ti 2Pb is formed atthe surface in areaction zone, destroying the sound metal. Also, problably by andissolution-precipitation process, crystals of an other intermetallic compound,Ti 3Pb 2 , deposit on the Ti surface. Both compounds have not been reported in theliterature. Also solubilities of Ti in lead respectively Pb-17Li and of Pb in solidTi could be determined, improving the knowledge of the Ti-Pb system.

List of experiments with titanium.Header line temperatures 0(, other numbers hoursof exposure time.

1 400 450 475 500 535 550 600 645 r62 · · · · · ·

· 140 · 167 · ·· · 235 262 261 · ·

· · 476 · 490 450 450

· 476 · 490 450550 · · 650 634 · ·

· 835 · 907 792 · · ·1176 · 1032 · · · 1100 ·2016 · 2080 · 2415 · ·3048 · · · · · ·

· 3790 · · ·· 6200 · · I · · I

Fig. Ti-lSolubility of Ti in Pb-17Li600 500 oe 400

10001------:---_-_~-_+:..;

+ =Solubilityin lead

100

10 : .. _. :. __ .

1.00 1.201000/T

1.40

oe400

Fig. Ti-2Difussion coefficient

700 600 5001[ -8.....--:1----+--..,---+:---..,....---1.1------,

1[-9- ---'-

1[-11 ---

1.601.401.201[-12-t------.-------r-------f

1.00

1000/T

14

Fig. Ti-3Solubility of Pb in Titanium

600 500 Oe 400

+ Exp. with lead

•• - - - •• - I •••• _ •• __ • • ~ •• _10

0.1 .

1.00 1.20lOOO/T

1.40

Fig. Ti-4Phase diagram Ti-Pb from ref.(61)

1000-r------------------,

I<> this work I

•••••••••••------------'''------4

<><>

.....

--------..--............

.0>

uoQ)

3800.....,Jctl~Q)

~

SQ)

E--o600

82 4 6at.% Pb in Ti

400t----....,...----,r----..,---~- ......o

15

Fi~. Ti-5Mieroprobe seon of the somple Fig.Ti-7, cut A-B.(1 +3 = reoetion zones, 2 Ti foil, 4 = erystol)

3....----------------------,

~1: 2 3:

1 1 'IA ~ Ilample :

4crYlltal

EJEJ

200150100mikrons

50

II

--r- .. -

' ........ -- -'- - , . .-.. _.~. ~ -. :I

:."'"61•t'i.....~Ii'r-- :...-;:

:\_~---~ i1 \o-l---==~---~--~-====___I

o

.-

16

Fi -6

Ti metal.Upper picture : without exposure to Pb-17Li.Lower picture : after 476 hours at 500 oe in Pb-17Li.Reaction zone only.

11

F'i -7

Ti metal after 634 hours at 550 oe in Pb-17li.Reaction zone and formed crystals.

Fig. Ti-8

•the Ti surface exposure to Pb-17Li.

19

Fig. Ti-9Formation of crystals

600 500 Oe 400100.......-----f---:------+----:---h

:0.' - _.. - ;(>0-· _ - - - _..

o

.......... _..... - ~O····· - _.....

o

1.401.201000/T

0.1-1------r--------r----'1.00

85

VanadiulII========

Vanadi um and its a11 oys were proposed as structura1 materi a1 for fusi on reactorb1ankets (16). Vanadium was also proposed as a getter meta1 for tritium extractionfrom the eutectic (65). A 1arge number of tests were performed with this metal.

Not much has been pub1ished for the systems V-li or V-Pb (66). No phase diagramsare availab1e from experimental data. A ca1cu1ated li-V phase diagram shows noreaction products.

1. Experiments

The tab1e shows the performed experiments. Vanadium was obtained from Goodfe110wwith 99.8+ % and from Ke1pin (leimen, Germany) with 99.7% qua1ity. The meta1 wascl eaned and vacuum annea1ed at 8ee oe for severa1 hours. There were some experi­ments in a1umina crucib1es. But in this report on1y resu1ts obtained with Mo cruci­b1es are used.

2. Results

First results on vanadium were a1ready reported in 1988 (6e). There were no re­marks about wetting of vanadium in 1aboratory notes. We assume therefore wetting inall experiments. This would be in agreement with observations for hydrogen extrac­tion where no incubation time was found after immersing the vanadium into the mol­ten liquid (65).

Dissolution

Fig.V-l shows the results of a11 experiments. It can be seen from Fig.V-2 thatthe dissolution does not stop after the initial period. Mass transfer to the cruci­ble wall is observed. The steady state dissolution rate after the initial period,Fig.V-3 , is given by

ln R = 2.37 - 5431/T

(R in g/m2*d). The heat of activation for the dissolution process is -45 kJ/mol,smaller than found for most other metals.

From Fig.V-3 it is clear that corrosion experiments can not find a weight loss withsmall samples, even if dissolution rates in non-static systems are higher. On thecontrary, Borgstedt (16) found in the Pb-17li-loop PleOlO a weight gain because ofiron deposition. Brasunas (15) found no attack of lead on vanadium after lee hoursat 76e oe. Ali Khan (22) observed areaction between vanadium and lead at 95e andllee oe after lee respectively 4e hours.

From constant dissolution rates and solubilities (below) diffusion coefficientswere calculated (Fig.V-4 ). The temperature dependence can be expressed by

ln 0 = -25.7 + 23ee/T

(0 in m2/s). The heat of diffusion is +19 kJ/mol. Such a sma11 value is expectedfor a diffusion process. But the value is positive as in case of beryllium. Probab­ly the used model is too simple. The average value between 4ee and 6ee oe is 1.5*le-18 m2/s.

Solubili of V in Pb-17li

As with other metals most of the dissolved V was transported by isothermal masstransfer to the Mo-crucible wall, the concentration in the bulk Pb-17Li remainedconstant. Fig.V-5 shows the solubility of V in Pb-17li. The function is given by

ln S (wppm) = 10.1 - 7730/T

The heat of dissolution is -64 kJ/mol.Only a few values are reported in the literature. They are included in Fig.V-5.

Al i-Khan (22) gi ves a 1i mi t of the sol ubi 1i ty in 1ead wi th 1ess than 10 wppm at1000 0C. But there was clearly dissolution as could be seen from metallographieinvestigations. A theoretical calculation of the solubility of V in Pb gives 72wppm at 600°C (10). But such calculated solubilities are very questionable(Part-I). Very low solubilities in lithium of less than 1 wppm at 727°C and lessthan 3 wppm at 1004 0C (66) are considered as not reliable. (The authors assumethat the data gi ven in ref. (66) are mi xed by mi stake). Beskorovai nyi (67) gi ves asolubility of V in Li with less than or equal to (?) 30 wppm at 1000 0C. Rumbaut(68) mentioned strong corrosion of V-Cr alloys in Li at 700°C and preferentialleaching of V. Therefore, solubility of V should be higher than that of Cr. Howe­ver, no Cr values were reported (6,8).

Solubility of Pb in V

The solubility of lead in vanadium was found to be less than 0.02 wt.% at alltemperatures. Also Li could not be detected in vanadium samples. Because someinformations are lost, no solubility limits can be given.

Reaction zone

A very thin chemical reaction zone was found during the electrolytical purificationprocess of samples from 550 to 650°C experiments. Fig.V-6 shows that the thicknessis proportional to the exposure time. The growth of this zone can be described by

d = 1.4*10-5 * t + 0.001

(d in microns, t in hours exposure time). That means the reaction zone thickness isonly 0.01 microns after 1000 hours.

Samples exposed up to 500 oe. show a surface layer if examined by metallography.These layers are not seen in experiments at higher temperature. (Fig.V-7:.Fig.V-8). Such surface layers were described e.g. by Borgstedt for V-3Ti-Si inflowing Pb-17li after 1058 hours at 550°C (16). The layer in his experiment was 55mi crons th i ck and very hard. It was formed because of pi ck up of non-metall i celements from the molten metal. A si mi 1ar very hard 1ayer was found by Brasunas(15) after 100 hours at 1000 0C in lead. The layer formed in our static experi­ments may have been formed by the same mechanism. But this layer was very thin anddid not prevent the dissolution of vanadium. (A thin hard surface layer was obser­ved on vanadium, exposed for 2500 hours at 450°C in flowing Pb-17li in loop TRITEX(69)).

Corrosion, visible by metallography, starts at 600°C, Fig.V-8 . But only unetchedsamples show this. Ali Khan (22) found strong corrosion of vanadium in lead at 950(100h) and 1110 0C (20h) and the formation of large grains. This was not observedin our experiments with lower temperatures. He also observed the formation of areaction layer on vanadium.

81

3. Summary

Vanadium dissolves in molten Pb-17li. but the solubility is low. Also. dissolutionrates are low. Because of isothermal mass transfer to Mo. dissolution does notstop at saturation. Solubilities of vanadium in molten Pb-17li and diffusion coef­ficients were determined. Besides a surface layer at lower temperatures. probablycaused by pick up of non-metallic elements from the eutectic and/or atmosphere. avery thin chemical reaction zone was found at higher temperatures.

Table: list of experiments with vanadium.Header line temperatures 6e. other numbers hoursof exposure time.

1 400 460 505 550 600 645 r167261 252

319319476 476 450

668907907 852

1130 119012201250

1460 12502000 208620002300

2720

88

Fie;. 1Dissolution of vanadium in Pb-1 7Li

(wld=weight loss data)

50002000

···;···0··;···. .'\}:

200

()

,~+ '~. .

....... : :';2: .

: + ~

: : .$........................... :.~ ~ .. T .

(): :() .

~ '7~ €l3

0.01+--------"""-----r---------.......100

10

..-Ne

............t:lIl

'-'

......l::;::10e<:tl

'" 0.1CL>::>-......00000.....'"

Fig. V-2Dissolution of vanadium in Pb-17Li

1.60,,.---------:--------:-----:-------,

" t.'. ,

.. __ .. ' T .1.20 , ;"" . .

~ 550 oC

+ 600 oC

:".'."

'"'"'"'", '",'".•.... : : ;;;,/; .

'"· '" '".+ '"· '". . '"· '".",

.. : ''Y' ,..."-:: .: "'~:. '": ~ ",'". '"

:",./

0.40

-C\l

S..............0.80Q.()---

300020001000O.OO-f---------r-------r---------I

ohours

Fig. V-3Steady state dissolution rate of vanadium

600 500 400 oC1E-1.......-i:'------I.~-:---__+·--_:__--+_:-----,

.... .... .... .~.... .<:::;> \>.... :0.....

~ <> .<> . .... .....- ~S .~.~ ~ - - .. <>........ .. .

<> <> ~<>

1.601.401E-3-t-------r,--------"r--------I

1.00 1.201000/T

oC

Fig. V-4Difussion coefficient

700 600 500 4001E-8 -r-l:I-----l:I---:----f.----:----+:--,

. .1E-9 .........................•..............

--~ . ~-_.N 0 -_ .. --,g _ :.. '(>-' .;. 0-- ~. :-.': :"". -:- .~ ~ ~ .. - .Q

1E-11 - .... - . - i - .. - - . - .. - - .. - ', . . - - - - .. - .. - - - .

1.601.20 1.401000/T

1E-12.+------r-------r-------11.00

Fig. 5Solubility of vanadium in Pb-17Li

100 1000 800 600 500 400 oe

• Gum.(10);... calc.for Pb

_L' "..... ". "Besko.(67) " ..... equol or

Li "-," ' ,Iess thon" . '- - - -~ Khan (22)- ;..._ . - - - .. - •.. - . - ... ~ . - 1-----1, Pb . " : .•this work

. ,..... '

'Smith (66) ,:Li :

.80

lII(dl):0.1 ""'----r-------tr----"I""---_--_;_----I

1.60

Fig. V-6Chemical reaction zone on vanadium

550 to 650 oC

0.04 - - . - . - . - - - . - . . . - - - - .... - .. ~ - . - - .... - ·0- - -:\1

0.02

o

3000200010000.00+--~___.....----__r----__1

ohours

91

Fi

Vanadium after ge7 hours at see oe in Pb-17li.Thin gray surface layer in upper picture of an unetchedsample. After etching (as Nb). th1s layer 1s lost(lower picture).

92

g.V-8

Vanadium after1220 hours at 600 oe in Pb-17li.No surface layer. Week corrosion attack seen inetched sample (lower picture).

93

Tungsten========

Tungsten will not be considered as a structural material for liquid metal blankets.A few experiments were performed with W-crucibles. From the analysis of Pb-17Li itcan be conc1uded that the sol ubil ity of W in the eutecti cis 1ess than 1 wppm at600 Ge.

Alden (56) reports a solubility below 23 wppm at 1200 Ge. Ali Khan (22) found noattack at 950 Ge after 306 hours in lead, Brasunas (15) no attack at 1000 Ge after100 hours. Tungsten is like molybdenum very stable in the eutectic.

Our boundary value, as well as the value of Alden, would fit well in our functionfor Mo. Because the values are upper limits, actually solubilities are lower thanexpected from ref.(10).

Solubility of Zr in Pb-17li

Fig.Zr-2 shows solubilities of Zr in Pb-17li. The funetion is given by

ln S(wppm) = 11.2 - 3423/T

The heat of dissolution is only -28.5 kJ/mol. Zr solubilities are remarkably highif eompared to titanium, but the heat of dissolution is small. There is only onevalue published for the solubility of Zr in Pb : 12 wppm at 500 Be (54), extrapola­ted from solubilities in Bi-Pb. Our value from the funetion would be 873 wppm.Weeks (54) found for the same temperature in Bi a solubility of 1200 wppm. Diffe­rent behavi our in Pb and Bi was also found in the system wi th Ti. But with TiWeeks and our solubility values are not so mueh different.

Experiments with Ti-Zr alloys (below) showed proportionality between the mole frae­tion X(Zr) in the alloy and the equilibrium eoneentration in Pb-l7li. That meansehemieal aetivities in the metal determine solubilities, and not a formed eompoundas in ease of titanium. 9 A value of 250 wppm for 500 Be ean be evaluated from theseexperiments for X(Zr)=l. This is 3.5 times lower than ealeulated from the funetion,but still mueh higher than expeeted from (54).

From solubilities and steady state dissolution rates diffusion coefficients wereealeulated. They are shown in Fig.Zr-3 The values are in an expeeted range. HeArrhenius funetion is given by

ln D = -17.6 - 5220/T.

The heat of diffusion is -43 kJ/mol, rather high for a diffusion proeess.

Reaction zone

Metallographie examination of samples shows reaetion lones at the surfaee. Examplesare given in Fig.Zr-4 and Fig.Zr-5. As with Ti, reaetion lones grow thieker athigher temperature and with longer exposure time. The ehemieal analysis gives 18.8±1 at.% Pb. This would eorrespond to an intermetallie eompound Zr4Pb. Two interme­tall i e eompounds are 1i sted in the phase di agram (12) : ZrSPb3 and ZrsPb, thelatter with a question mark. We do not know if, as for the Ti reaetion lone thaton Zr has a homogenious eomposition. No mieroprobe analysis was done. Therefore,ZrsPb may also be possible.

Solubility of Pb in zirconium

There was no determiniation of the eoneentration of Pb in sound Zr. But as in easeof titanium, there is elearly a dissolution of Pb in Zr. The remaining metal afterremoving the layer of reaetion produets was very brittle. The effeet is deseribedin the literature (12) but not investigated in our work.

9 Otherwise a compound at the surface would have also Zr concentrations proportional toX(Zr). This, however, was not seen.

Hit

3. Summary

Like Ti, Zr cannot be used to extract tritium trom the molten mixture. Reactionzones of intermetallic Pb-Zr compound(s), probably Zr~b, are tormed on tlle surfaceuntil the whole sample is destroyed. Tlle solubility of Zr in the eutectic was foundvery high, with a low value for the heat of solution. From experiments with Ti-Zralloys, mud lower 'solubility'-values were derived than for pure zirconium. Theeffect will be discussed below with alloys. Diffusion coefficients were in an ex­pected range.

list of experiments with zirconiumHeader line temperatures oe, other numbers hoursexposure time.

1 400 450 500 535 550 600 645 r

· 62 · · · · ·· 140 · · · ·· 262 · 252 ·· 476 496 · · 450

· 476 · · ·· · · · 610 · ·· · 821 792 · 826 ·1110 1006 · 1100 1150

· 1320 1514 · · 1100 ·2110 1700 1826 · · · ·2470 2320 2000 · · · ·3013 · · ·· 3466 · · · · ·· · · · ·6200 · · · ·

162

Fig.Zr-lSteady state dissolution rate of Zirconium

600 500 400 oC

10 : ~ .

lass ofsound metal

0.1 . - . . . ..' ........ : , ""7 .

1.0 1.2 1.4IOOO/T

1.6

Fig. Zr-2Solubility of Zr in Pb-17Li600 500 0c 400

1OOOO.-----+-----:---+---~-_.......

1000 ....

1.401.20IOOO/T

100 +------...-------....------'1.00

103

oe400

Fig. Zr-3Difussion coefficient

700 600 5001[-8 .......-i.I-----l:I---;-----i-:----;----+_.---,

1[-9· - - . - - - . - - - . - - - : - - - - . - - - . - .. - - ; - .. - - - - - - ... - .

. . . . . . . . . . . . - - - - - - - .

:0<5"~ ....

1[-11 - - ..o 0~ ....

. - - - - - - .. - . -. .

1.601.401.201[-12+-----......----__.-------1

1.001000/T

U.l4

Fig.Zr-4

lr metal after 1320 hours at 450 oe in Pb-17li.lower picture:Reaction zone removed by etching with121ml "20 + 21ml HN03+ 6ml HFI.

11:'15

Fig.Zr-5

Zr metal after 1514 hours at 500°C in Pb-17li.Contact between reaction zone and sound metal partly lost.

121

Other al1oys. steels====================

Mo-Re alloys

These alloys were described before in chapter Rhenium. They are probably more stab­1ethan Mo-meta1 (59).

TZM

TZM (Plansee) is often used instead of pure Mo because of its better workability.It contains 0.5 wt.% Ti and 0.1 wt.% Zr in Mo. It was found as stable as Mo metal.After 1000 hours at 600 Ge Ti and Zr could be detected in the eutectic, but thevalues were only 0.1 wppm Ti and 6 wppm Zr. Because of the low dissolution rate,these elements were assumed not to be in equilibrium with the solution. There wasno reaction zone visible.

TZM can be used for Pb-17Li systems like Mo metal.

Austenitic stainless steels

Austenitic stainless steels with different composition (steel 4571, 4300 andothers) were tested at the beginning of the experimental program. Many corrosionexperiments were performed with austenitic steels. Ref.(72) is given as an example.All effects, especially leaching of Ni and the formation of a ferritic surfacelayer, were also found in our experiments. We have reported at a workshop (33)about a compound MnNi, found in thermal convect ion loops at the spot wi th lowesttemperature. In the meantime this compound was identified in other facilities undproposed to remove manganese from the eutectic (35). The compound could not be seenin isothermal compatibility tests. That means the solubility of this compound atexperimental temperatures is high.

Because there were no new effects no results will be given in this report.

122

Ferritic steel 4922

This is the steel of the TRITEX facility in our laboratory (38). The steel con­tains 12 % Cr, 0.5 % Mn, 0.3 % Ni, it is stabilized by 1 % Mo and 0.3 % V. Firstresults from static compatibility tests were published in 1988 (60).

Results

A too long time passed since the investigation of this steel, not all data are nowunderstandable any more. Generally observations are in agreement with the litera­ture for ferritic steels.

A two step dissolution is shown in Fig. 4922-1 with a higher dissolution rate atthe beginning as found for most metals in our experiments. The steady state disso­lution rate is given by

ln R = 4.06 - 5195/T

(R in g/m2*d). The heat of activation is -43 kJ/mol.

Chopra (74) found also for ferritic steels a faster dissolution at the beginning,but the mechanism may be different. Tortorelli (11) found only a linear weight losswith time while Borgstedt observed a long incubation time with much lower disslolu­tion rate than under steady state conditions (50).

Inspite of such odds, steady state dissolution rates are compared with corrosionrates of ferritic steels in flowing and static Pb-17U (Fig. 4922-2 ). Tas (72)found 1988 that all data fit between 2 1i nes, a factor of 28 apart. Newer val uesfrom Sannier (76) and Borgstedt (77) fit also between these lines. As expected forstatic tests our values are lower. A few values from Tortorelli (78) for staticcorrosion of HT-9 fit better to our function (values taken from ref.(6)).

In equilibrium at 550°C the eutectic contained 3 wppm Fe, 2 wppm Cr, 0.8 wppm Mnand 15 wppm Ni.

The equilibrium concentration for iron is higher than obtained with alpha-iron me­tal (see Fig. Fe-3). but still in the range of values from the literature. Also,dissolution rates are higher than for pure iron. In loop TRI TEX (69) particles withdifferent Fe : Cr ratios were found. Therefore we assume that in case of steel4922 not iron but Fe-er compounds are dissolved.

Using mole fractions X for chemical activities of the elements in steel and solubi­lity functions from Barker(8), 8 wppm Mn and 12 wppm Ni are expected in solution.The Ni value was found. Nickel is in an ideal solution in the steel, it is clearlyleached out. A nickel depleted zone should exist at the surface, but this was notseen (Fig. 4922-3 and 4 ). The Mn value is too low, pointing to non-ideal behaviourof this element in steel. No solubility function is reported for chromium. Sample(6) and Barker (8) found widely scattered data between 2 and 18 wppm, correspondingto 0.3 to 2.5 wppm are expected for experiments with steel 4922. The found value isin this range.

123

No 'metallographical' reaction zone is seen (Fig. 4922-3 and 4 ). At higher tempe­rature and longer exposure time, the roughness of the surface increases. This pro­cess probably continues with more dissolution, as seen by Borgstedt (58) for steelX18CrMoVNb 12 1 in PICOlO loop.

However, a 'chemical reaction zone' was found. During the eletrolytical purifica­tion process, more material is dissolved than from fresh (and equally heat treated)stee1. Such an effect was descri bed before for pure i ron and other metals. Typi­cally 8.85 microns were removed from a sample after 2800 hours at 550°C. Thisreaction zone is proportional to exposure time and thicker at higher temperatures.

Summary

The ferritic steel 1.4922 gets dissolved by an even dissolution process. The disso­lution rates are lower than seen with flowing Pb-17li, but higher than those ofpure iron. Also iron 'solubilities' from 4922 experiments are higher than obtainedwi th pure i ron.

No reaction zone can be seen with metallographical examinations. A thin chemicalreaction zone is identified. Here is clear leaching of Ni from the alloy. Nickelis in ideal solution in the steel. while manganese is not.II

11 More i nves t i gat ions were dane fram TRITEX samp1es. The res ults wi 11 be pub1i s hed inanather paper (69).

124

Fig. 4922-1

Dissolution of steel 4922 ot 550 oe

4- _ :. _ : V .

II

II

II

II

I

400030002000

hours1000

O+------r-----r-----......,-----lo

Fig.4922-2Corrosion of ferritic steels in Pb-17Li

600 500 400 oC100 .....-----..........-.---1----.,..-----tl------,

1.601.401.20

" .

10 '" \", ."80;'9.(77)~"" :

Tos(72), : ",,~os~72), '.., . "-

1 - : . '., - . , -:'Sönn".(76\ .: " '... "-

O. . "-, . "-

A 'V)RITEX:-:-:-.,..,.~O~..; ~ ~~._ ....

V : A Tort.(78)

. 1.49220.01 . - ___.... - . ; - - .. lhis work .

.~. .:~. iron. ~this work

------0.001+--------..------.=------11.00

'"'0

*C\1

S.............

b.()

0.1

1000/T

125

Fig.4922-3

Steel 4922after 1138 hours at 400°C in Pb-17li.Two positions.Etching with1100ml ethanol + 4g picric acid + Iml HCl I.

126

Fig.4922-4

Steel 4922after 1536 hours at 550 oe in Pb-17li.Two positions.Etching as before.

121

Mo coatings===========

Metallic coatings are widely used to protect less noble metals against corrosion inatmosphere or aqueous solutions. In liquid metal systems coatings are of limitedvalue. Instead of easy to understand electrochemistry, more complicated dissolutionprocesses and sol ubil iti es are important. Furthermore , because of the usua11 y ap­plied high temperatures, mutual diffusion of the metals has to be considered.

Mo coatings on stainless steel 316 were studied in connection with sodium cooledreactors (7S). While the corrosion rate of steels in sodium at 7ee 0C was reduced,poor adherency of the Mo layer was found. Blisters were formed and the coatingpeeled off.

Asher (36) has tested Mo and other coatings on steels at 7ee 0C in liquid lead.With the exception of Al, no coating was successful to protect the steel againstcorrosion. There was always an 'uptake of lead' by the samples, the authors descri­be the effect of 'lifting, spalling and flaking' of the coating.

In liquid lead systems at high temperatures, Mo coatings on stainless steel werestudied by Block (37). The main problem besides adherency was the diffusion ofnickel trough the Mo layer. This is not only caused by the high solubilty of Ni inlead. With high Ni concentration in the steel, well formed crystals of Ni grow atthe surface after only lee h at llee 0C.

1. Experiments

In spite of thi s di scouragi ng results from the 1i terature, some Mo-coated geHermetals were investigated. As discussed before, Metals like Y, Ti or Zr cannot beused directly to extract tritium from the molten mixture. The metals were coated byS mikron Mo by different companys: Pansee, Goodfe11ow, Kammerer, mostly by PVDtechniques. The goal was to protect the metals against corrosion, but keep theresistance against tritium permeation low. The results with different coating tech­niques were identical. This is in agreement with observations by Asher (36). There­fore, companys or coating techniques are not mentioned any more in this report.

Experiments were performed mainly at see and 550 oe with exposure times up to 520ehours.

2. Results

Yttrium - 5 mikron Mo

It was not possible to obtain good quality Mo coating on Y metal. Always the Moski n was loose after cutti ng or bendi ng of samp1e pi eces. The problem i s causedprobably by an oxide layer at the Y surface. Therefore, not many experiments wereperformed. Mo did not protect Y, dissolution rates were very high and comparablewith uncoated metal. e.g. samples were completely dissolved after only lee hours atsee 0C, leaving the Mo-skin behind.

128

um - 5 mikron Mo

Fig.Ti 1 gives an example for the dissolution of titanium without and with a Moeoating. In the figure, the formed erystals are not ineluded. It ean be seen thatTi is protected at 5eB BC for nearly 1000 hours. But then the dissolution rate isthe same as without eoating. A metallographie examination, Fig.Ti/Mo-2 , shows thatafter extended exposure time the Mo is partly detaehed. the Ti metal shows a reae­tion zone as seen on pure Ti and the same kind of erystals are formed on thesurfaee of Mo as before on Ti.

DiscussionMo is proteeting Ti only for a short time. Then Ti is diffusing through the Mo

eoating, dissolution rate and formation rate of erystals are the same as for uneoa­ted Ti. The protect i ng time may be long enough at lower temperature for usi ngeoated Ti in Pb-llLi. It ean however not be used as a tritium getter beeause thetritium recovery from the getter metal has to be done at least at 6B0 BC • At thistemperature, however, Ti with or without coating is destroyed within a short timebecause also lead is diffusing through molybdenum.

Al10y Beta-3 - 5 mikron Mo

The behaviour of Mo-coated alloy Beta-3 was identieal to the behaviour of Mo-coa­ted titanium. Therefore, no results are given in this report

Zirconium - 5 mikron Mo

Fig.Zr/Mo-l shows the dissolution of Zr without and with a Mo coating at 500 BC,Fig.Zr/Mo-2 at 550 BC. There is clearly a protection effeet at the lower tempera­ture. But at 550 BC the behaviour is similar to titanium : Zr is protected only fora short time; then the dissolution rate is the same as without coating. A metallo­graphie examination was done for one sample. Fig.Zr/Mo-3 shows only a thin Mo layerremaining. We do not know where the missing Mo was gone, e.g. if it was dissolvedin the Zr metal. No more investigations were done.

Discl.IssionAt 500 BC, Zr ean be protected for some time by coating with Mo. The behaviour at

higher temperatures is similar to titanium : after an initial time the same disso­1uti on rate i s found as for uneoated lr. Zr can not be used as a trit i um getterbeeause the triti um recovery has to be done above at 600 BC • At th i s temperatureZr with or without eoating is destroyed within a short time.

3. Summary

The investigated thin Mo eoatings proteet Ti and Zr only for a short time againsteorrosion. As the uneoated metals also the coated metals eannot be used as gettersfor tritium extraetion because the tritium recovery has to be done at too hightemperatures.

10- ...

129

Fig. Ti/Mo-lDissolution of titanium at 500 oe

without and with Mo-coating30.,..--------;-------:---------,

:Ti/~20 , , .

////

I'I'

I'I'

: I',;

...... ,I, .....I' '

I'

1000

hours2000 3000

Fig.Ti

Mo ng on Ti after1000 hours at see oe in Pb-17Li.Two positions.

131

Fig. Zr/Mo-lDissolution of zirconium at 500 oC

without and with Mo-coating

~100 - - .. - ... - - ... - .. -Zr- -. --..... ----.. -. -.. -. --

.. -: _.. Zr-Mo· . _.

600040002000hours

O-l-~------;r----------------l

o

Fig. Zr/Mo-2Dissolution of zirconium at 550 oC

without and with Mo-coating100..,-----r--r-------.,.---------.

Zr

. -\:1- ...

IIIIIIIII

__ .0_ ... ;_

o

Zr-Mo

150010005000-1---..4-----;------;--------1

ohours

132

Fi g. LV' ,IMI'!I-

Mo coating on lrSOO hours at see oe in Pb-17Li.Lower cture etched as zirconium.

20 m

133

Part-III : Description of experiments

1. Used Materials and preparation 135

1.1 Eutectic mixture Pb-17li 1351.2 Crucibles and boats 1361.3 Sample materials 136

2. facility and experiments 137

2.1 Initial Setup 1372.2 Exposure 1492.3 End of exposure 149

3. Chemistry and analysis 141

3.1 Purification of samples and crucib les 141- Electrolytic method

3.2 Preparation for analysis 144- Extractor - Chemistry

3.3 Analysis 146- Chemical Analysis- Metallogaphy and other investigations

4. Evaluation of results 147

4.1 Units and definitions 148

135

1. Used Materials and preparation

1.1 Eutectic mixture Pb-17li

The eutectic mixture Pb-17Li was obtained in 1986 from Metallgesellschaft. Frank­furt. It was shipped in a very coarse way. the surface of the ingots had to beremoved mechanically. After this all handling of the eutectic was done inside of anargon fill ed glove box. Oxygen and humi dity concentrat ions in the box atmospherewere measured contineously and kept below Ippm. Nitrogen was measured occasionallyand found up to 500 ppm.

Further purification of the eutectic was done by several times remelting. alwaysremoving floating crusts. There was no knowledge of the problem of segregation (7)at the time of this treatment. Therefore. the Li concentration was different ineach experi ment. However. ana lyzi ng the results in respect of Li concentrat ionsshowed no influence within the error range of data.

Chemical analysis was done mainly by ICP-AES (below). The following concentrationsof elements were found :

Table 1 : Impurities in eutectic mixture Pb-17Li.Except for Li. all values are given in wppm.

Li 0.62 to 0.77 wt.%(15.7 to 18.8 at.%)

Na 28 Be 0.042K 2.0 Mg 0.45

Ca 1. 55Sr 0.02Ba 0.e1

-----------------------------------------B 0.88 (?) Sn 0.07Al 0.34 (?) As <0.4Ga <0.5 Sb 0.06In <0.5 Bi 53Tl 1.9

Se <0.2Sc <0.5 Te 0.02y <0.01

-----------------------------------------Fe 4.3 Cu 0.22Cr 0.25 Ag 6.1Ni 0.18 Zn 0.52Mn 0.042 Cd 1.5Co < 0.5 Hg <0.4Mo <0.05V 0.01

136

Other elements were not determined. After crust removal during remelting, the eu­tectic was probably saturated with oxygen and nitrogen. Solubilities of these ele­ments are very low (71,82) and further reduced during vacuum degassing (below).Also some used sample metals were getters for these gases. 12

Brewer (25) points to a special problem when using molybdenum : the formation ofvery stable Mo-carbides. Carben could be transported e.g. from iron samples to Mo.The role of carbon in Pb-17Li systems is not yet studied. Borgstedt assumes a verylow solubility and, therefore, only negligible transport (3).

1.2 Crucibles and beats

Most experiments were done with Mo-crucibles. They were obtained from Plansee, pureMo with <0.5 % impurities. The size was 25 mm diameter, 40 mm high with 0.5 mmwall. Crucibles were chemically etched and vacuum degassed at 700 GC for severalhours at a final pressure below 10-~ mbar.

Only a few tests were done with crucibles of iron, tungsten or alumina. The metalswere purified as described in Part-lI. Alumina crucibles (Al-23, Friedrichsfeld)were cleaned with boiling HN0 3 , heated in air to 900 GC to remove all organicmaterial and vacuum degassed at 1000 GC for 10 hours to a final pressure below 10-4

mbar. They were cooled under purified argon.

Al z03 boats were used at the end of exposure (below). Boats (Friedrichsfeld) weretreated as crucibles. 13

1.3 Sample materials

Sample materials were obtained from different companies. They were chemically clea­ned and vacuum degassed for several hours until a final pressure below 10-4 mbar.Chemi stry as we11 as the temperature of the vacuum treatment, was di fferent foreach metal. Details are given in Part-lI. Some informations are lost. All volatileimpurities were removed by this treatment. Furthermore for most of the investigatedmeta1s oxygen and nitrogen from the surface di sso1ved in the meta 1, 1eavi ng thebest possible surface for the experiments. Beryllium metal was mechanically polis­hed in some tests.

12 The color of samples and the appearence of the eutectic after exposure was an indicationfor trace amounts of 0 and N in an experiment.

13 Because in some tests al umi na reacted with Li from the eutect i c, on 1y a few experi mentswere done with alumina crucibles. The results are not included in this report.

2. 11; and

131

Mo-crucibles with pes of sample metals in 17Li were heated in stainlesssteel capsules. Figure 1 shows a drawing of a capsule, Figura :2 CI photo of thearrangement of crucibles. Because of heat reflectors and radiation shields thetemperature across the crucible chamber was constant within ±0.50C. This should begood enough to avoid convection flowS. 14

2.1 Initial Setl.lp

All handling of the eutectic was done inside of the argon filled glove box.

Crucibles were filled with Pb-l7li up to 20 mm high. Ihe wetted Mo-surface wasabout 15 cm2, 6e to ge grams of eutectic were used.

Then sheets of sample were fixed in the molten metal with stripes of Mo as seen infigl.lre 2. The sampl e surface in contact with the eutecti c was 3 to 8 cm2 • Thesurface to volume ratio, therefore, was similar to the crucible. Any contact bet­ween sample metal and crucible below the liquid metal surface was avoided. Otherwi­se often areaction between the two metals caused a failure of the crucible. Suchan effect is well known from sodium systems and observed by Khan with steelsand high melting metals in lead.

Still inside of the glove box and after solidificaton of the eutectic, crucibleswere set into the holder and the capsule was closed. Up to 6 crucibles were placedin one capsul e, often with different metals to be studi ed. The capsul e was takenfrom the box to the laboratory.

The next step after a He-leak test was the degassing of the Pb-17Li. The capsulewas evacuated and heated to the experimental temperature. Heating up had to be donevery slowly to avoid bubble release of gases and splashing of the eutectic. Thecapsule was kept for several hours at apressure below 10-4 mbar. Then the capsulewas fi 11 ed with Ar-6. e, further purifi ed near the entrance by Oxi sorb-S (MesserGriesheim). The procedure of vacuum degassing should also help wetting of samples,but there are no observations reported in laboratory notes.

14 Brasunas (15) mentioned that in crucibles with lead thermal convection can be expected attemperature di fferences of ± 2°C.

138

Figure 1: Facility for static compatibility tests

argon-.

inlet

vacuum-system~

argon ouilet

radiator

heat reflectors

::JJ.:EC._---crucibles

heater

139

Figure 2: Arrangement of crucibles

into a e heating. to 6 cruci es with different e materialswere in one capsule. During heating at constant temperature a flow of pure argonpassed through capsule.

After the exposure, crucibles were taken out from the capsule in the glove box.They were reheated und the samples extracted. The Pb-I1Li was poured into aluminadishes. The eutectic was dissolved in stoichiometric amounts of nitric acid, usinga special extraction technique. The solution was analyzed for dissolved metals byICP-AES. Remaining Pb-17Li at the crucible wall was also dissolved and analyzed.Finally remaining eutectic from the sample surface was dissolved and analyzed. Thiswas mostly done by an electrolytic technique. But depending on the material alsoother different chemical methods were used. There was a weight control of samplesas well as metallographical examination. For some metals fractions of the samplewere dissolved and analyzed for Li and Pb.

3. Basic considerations

A good description of all effects in connection with static:: c::ompatibility testsof materials in liquid metals is given by Brasunas (15). The limited value ofresults from static tests for dynamic and non-isothermal systems is discussed. Ho­wever, isothermal static tests contribute considerably to the knowledge of materialbehaviour. Loops are usually constructed from a material different from the sample.The surface ratio sample to loop is very small. The liquid metal becomes saturatedby not wanted metals, and mass transfer of structural materials to the sample sur­face is possible (16). Furthermore different elements in solution influence solu­bilities.

The driving force for the dissolution is the different chemical activity in a sam­ple metal and Pb-llLi. This activity is one at the metal surface, respectivelyequal to activities of metals in alloys. At equilibrium (saturation) the chemicalactivity is the same in the liquid meta1. In most cases, dissolution rates arecontrolled by the mobility of atoms in the liquid, that means by diffusion in theeutectic, but other rate controlling steps are also possible (11). The dissolutionprocess can be divided into the following steps

3.1 Wetting

Wetting of the sample surface by the liquid metal is the first step. The physicaldefinition of wetting does not help mud for the understanding of liquid metalcorrosion. Also it is not possible to see the shape of droplets of molten eutecticwithout special equipment. The practical definition as used in this report is

A wetted surface after removing from the molten eutectic at 350 to 400 GCretains a film of 30 to 100 microns of liquid metal equivalent to 30to 100 mg/cm2 • Non-wetted surfaces are clean with droplets.

Often a gap between a non-wetted solid surface and the liquid is assumed, preven­ting the contact and dissolution. Sometimes it is observed that corrosion startsafter a long incubation time. e.g. Borgstedt (17) found this effect for MANET-steelin Pb-I1Li. It was assumed that a thin oxide layer on the steel prevented wetting.

146

2.2

A number of investigated metals and alloys are getters for oxygen und nitrogen. Theatmosphere in the capsule was therefore strictly controlled. It was found better tohave an argon flow of 10 to 20 cm3jmin at an overpressure of 100 mbar than a sealedcapsule. The gas leaving the capsule was analyzed from time to time with an elec­trochemical oxygen meter. The oxygen concentration was always below 1 ppm. Alsohumidity and nitrogen concentrations were below lppm.

An additional control for the purity of the atmosphere was the appearance of Pb­l7Li and sample surfaces after exposure. Sometimes a slight tarnish was seen atsamples and a haze of oxides at the eutectic. Such oxide layers were extremely thin(86).15 In most cases. however. samples and eutectic were metallic clean.

Capsules were kept during exposure time at one temperature. constant within ± l8e.However. capsules were cooled sometimes to room temperature to exchange crucibles(glove box). This should have no influence on results.

2.3 End of exposure

After the exposure time capsules were cooled quickly to room temperature. This wasno quenching but it took only about 5 minutes to solidification. Therefore nomajor segregati on effects are expected. Then the capsul e was opend in the argonglove box and the crucibles were taken out. The eutectic mixture was remolten on ahot plate at about 350 8C. After removing the sample it was poured into an aluminaboat. Because of the fast cooling and short contact time of Pb-17li with the alumi­na. no uptake of Li. Pb or other metals was observed. There was no wett i ng. theeutectic could be taken away after solidification for analysis.

15 Because the sma 11 total amounts of 0 and N, propert i es of samp1e metals were not i nfl uen­ced.

141

3. Chemistry and analysis

Samples, crucible and Pb-17li had to be prepared for analysis.

3.1 Purification of samples and crucibles

After removing a sample from the eutectic and pouring out the crucible at 350 oe,4 to 8 mg/cm2 remained at sample and crucible surfaces, corresponding to a layer of4 to 8 microns. However, there was not always a homogenious layer of Pb-17li at thesurface. Depending on wetting, small droplets could remain, making the purificationdifficult. In the crucible often up to one gram remained in corners. This adheringPb-17li had to be dissolved and analyzed for the sample metal.

The purification of samples is a problem with liquid metal corrosion experiments inliquid metals. Especially the removal of lead and Pb-17U is difficult withoutattacking the sample surface.

Normal wet chemistry, e.g. dissolving adhering eutectic in acids, is in most casestoo aggressive. Only a few metals, when the electrolysis (below) was not possible,were treated by this way : Pb-17U was dissolved in a mixture of 2% acetic acidwi th 1ess than 1% H202 at 50 to 80 oe.

Often in corrosion experiments with Pb or Pb-17li, samples are washed with moltensodium (17,50) or lithium (11,85). In our experiments we would have to use extre­mely clean Na or U and dissolve it for analysis. To avoid handling of largeamounts of alkali metals, this method was not tested.

Attempts were made to remove Pb-17li by dissolving in mercury (36). Hg can easilybe purified by destillation. At room temperature about 400 mg/cm2 eutectic can bedissolved within one hour in 40 ml stirred mercury. Under the same conditions lessthan 2*10-~mg/cm2*h of metals like V, Nb, Ti, Zr, Hf, Mo or stainless steel getdissolved. Metals like uranium or yttrium dissolve or disintegrate in Hg.

The mercury method is complicated. Because of lithium it must be done in an inertatmosphere. Otherwise Li is oxidized and a black foam is floating on the surface.An attack of sample surfaces cannot be excluded. And finally the determination oftrace amounts of metals in mercury is difficult. This method was therefore usedonly at the beginning of the experimental program to get clean samples for metallo­graphy.

El c method

142

With a few exceptions, adhering eutectic was dissolved under extremely mild condi­tions by an electrolytic method:

The sample was used as anode, Cu-covered platinum as cathode.In most cases the electrolyte was a 10 % solution ofammonium-acetate. The potential of the cell was kept as low aspossible, usually between 0.1 to 0.3 Volt. The cell currentwas measured to get information about the progress of dissolutionand a possible attack to the base metal. The potential was highenough at the anode to dissolve lithium and lead. Standardpotentials of the other metals are much higher.

A special technique had to be applied for the cathode. Lead was mainly deposited atthe cathode, together with dissolved sample metal. It reacted there with platinumand could not be dissolved for analysis. Therefore Cu-plated Pt was used, the Cudissolved after the electrolysis in nitric acid, the solution analyzed.

Figure 3 shows a typical time-current curve for steel 4922. The first sharp drop inthe curve indicates the end of Pb-l7U dissolution. Then the 'chemical reactionzone' is dissolved, followed again by a sharp drop of the current to values below0.1 mA.16 Of course, such 'ideal' curves were only found for well wetted sampleswith smooth layers of eutectic at the surface. Visual control during the electroly­sis was important to detect droplets or thicker layers at parts of the sample orcrucible.

Sometimes problems occured because of the very low voltage and copper di ssolvedfrom the cathode ! (Reverse flow of current.) In this case results were not usedfor data evaluations in Part-lI.

The electrolytic purification could be used for nearly all metals and alloys. U andY could not be treated because of too high dissolution rates. Also alpha-Fe getsdissolved to a larger extent. The dissolution rate of Mo, 0.005 mg/h, was justsmall enough to use the method for crucible purification.

16 In many cases, the 'chemical reaction zone' was too thin to be seen in this curves. In thiscases it was identified by chemical analysis of the solutions.

143

Figure 3: Electrolytic purification of steel 4922.0.1 Volt, 10% ammonium acetate solution.Exposure of sample 3000 hours at 450°C.

III

15010050

•••••••.•••..• ! ••• - ••••••••• - .' - - •

~<>. : :

'0-0 -<7 --0- :--0- ............... ; ·0···········:··············

~

0.1

0.01 f---------.------..,.--------l

ominutes

144

3.2

The solutions from the electrolysis could be used directly for ICP-AES analysis.

The bulk Pb-l7U, 60 to 90 grams, taken from the alumini! boat, was dissolved innitric acid. For this a special extraction technique was developed, allowing touse nearly stoichiometric amounts of acid. This was an important item because theacid contains trace amounts of metals. To minimize the amount of acid, therefore,means to minimize background effects.

Extractor - Chemistry

Since about 1985 the extractor was used. Figure 4 shows a photo of three extrac­tors operating in parallel. While always clean recondensed nitric acid is dissol­ving the eutectic, lead nitrate is precipitating in the flask below. Only a smallexcess of acid is required. If areaction stopped, water and/or acid could be addedat the top without opening the apparatus.

Unfortunately in most experiments, excess nitric acid was evaporated after dissolu­tion of the eutectic, Pb(N03)2 dissolved in water, and lead precipitated as PbS04 byaddi ng dil uted sulfuri c ad d. Then the sol ut i on was adj usted to 1 1iter. PbS04

settled within a few hours, the solution could be analyzed. No metals precipitatedwith PbS04 but the adjustment to 1 liter caused reduced sensitivity.

A modified lead-analysis procedure was applied in some cases to reduce detectionlimits: Excess acid with the dissolved elements were taken from the Pb(N03)2 slur­ry, the latter washed with concentrated nitric acid. The combined acids were analy­zed. By thi 5 way about 100 ml of sol uti on coul d be measured i nstead of 1 1i ter,improving the detection limit by a factor of ten. (Only by this techique solubili­ties for Mo, Nb and Ta could be obtained.)

145

Figure 4: Extractors for dissol ng larger amountsof Pb-17li.

146

3.3 Analysis

Chemical Analysis

Analytical wet chemistry was in most cases chemistry with very low concentrationsbut with high concentrations of Pb and Li as matrix.

Allsolutions were analyzed by ICP-AES. using an ICP-6500 and later a PLASMA-2 ofPerkin Elmer. The matrix of many solutions was crucial for ICP-AES : up to 30 % HNO3 and 3 to 10 grams/l i ter Li and Pb each. But measurements up to 20 g/l Pb-17L iwere possible without 1055 of sensitivity, using appropriate plasma conditions.

Bl anks and standard sol ut ions were produced by di sso1vi ng appropri ate amounts ofPb-6.0 (Baker, Hereaus) by the same procedure. adding Li and desired elements fromMerck or Spex ICP-standard solutions at the beginning of the dissolution process.For metals near the detection limit an addition method was applied.

Metallogaphy and other investigations

Many metallogaphic samples were prepared during the experimental program in ourlaboratory. The semi-automatic machines 'Planopol' with 'Pedemax' (Struers) wereused. Most of the good pictures of Part-lI, however. were prepared by Mrs.Echtle.'Institut für Werkstoffkunde I, Universität Karlsruhe.' Microprobe analysis wasdone in the 'Institut für Materi alforschung' and the 'Institut für Mi krostruktur­technik', both Forschungszentrum Karlsruhe.

141

4. Evaluation of results

For each experiment three values for the dissolved metal were obtained

- amount of dissolved metal in the bulk Pb-17li,- amount of dissolved metal deposited at the crucible wall,- amount of dissolved metal from sample cleaning.

From the first kind of data solubilities were derived. If the concentration in thebulk Pb-17U was constant with time, saturation was attained. We assume that thediffusion profile across the crucible in case of isothermal mass transfer will havenot much influence on results: saturation of most of the eutectic, lower concentra­tion only near the crucible surface. There are no more informations about thisprofile from the experiments. From comparison with' literature values for some ele­ments, however, we think that the found solubilities are reliable values.

Dissolution rates were obtained from the amount dissolved in the eutectic and theamount from the crucible walls. This sum can be compared to weight-loss data fromcorrosion experiments. The dissolution rates became constant after an initial time.This 'steady state' dissolution rates were used in this report to describe thebehaviour of metals. Most often the dissolution was faster in the inital period. Ina few cases, however, an incubation period was seen. Not many da ta points exist forthe initial period; this period, therefore, is not discussed in this paper.

Even if steady state dissolution rates are of limited value for non-isothermal andloop systems, they allow at least to compare different metals with each other andto help in prediction of corrosion rates.

The amount dissolved during sample cleaning gives information about a chemicalreaction zone at the surface. 'lass of sound metal' would be the sum of all threevalues, taking into account also formed reaction layers.

In the case of isothermal mass transfer the First Fick's law is valid and diffusioncaefficients can be calculated.This is discussed in detail in Part-I.

148

4.1 Units and definitions

ln

temperature

logarithm are always natural logarithm on base 'e'

eentigrade GC, in Arrhenius funetions K

time hours or days

gas pressure mbar

mole fraetion x

solubilities

dissolutionrates

Diffusioneoeffieients

temperaturedependenees

reaetionzones

wppm. Only in some eases appm are used. Beeause of loweoneentrations, appm can be approximated from wppm bylappm = wppm * 176 IMI. 176 is themolekular weight of the euteetie with 15.8 at.% Li,Mthe atomie weight of the dissolved metal.

g/m2*d.Steady state dissolution rates are obtained afteran initial period. During the initial period dissolutionis most often faster, but an ineubation time was alsoseen. Dividing the values by the density in g/em3

gives mikrons/day.

Weight 10ss as often given from eorrosion experimentsean be ealeulated, while lass of sound metal needsalso informations about surfaee reaetion zones.

m2Js. They are derived from the First Fieks lawwith the assumption of Co«Cs at thedeposition surfaee. Diffusion eoeffieients wereealeulated only if solubilities were measuredat the same temperature.

of thermodynamie data are given in form of theArrhenius funetion:

ln Y = A + BIT

The heat of the proeess is given in kJ/mol.

Two kinds of ehemieal reaetion zones are found for somemetals. Chemieal eleaning of samples gives informationabout a thin reaetive surfaee layer. Metallographieexamination shows thieker modified surfaee layers.

Abbreviations for experimental faeilities

TCl thermal eonveetion loopFCl foreed eonveetion loopBE bateh type experiment, sometimes stirred

Appendix

References

list of companies

Acknowledgement

149

151

159

161